An Introduction to Failure Analysis for Metallurgical Engineers

CONTENTS
FORWARD PROCEDURE CASE STUDIES Introduction to Case Studies 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 of 316L Reducer Failure APPENDIX 1: EXAMPLE QUESTIONNAIRE Bibliography

FORWARD

PROCEDURE

To increase the odds of completing a conclusive failure analysis while at the same time saving time and money, investigations should be carried out using a systemic approach similar to that outlined in Figure P.1. It is important to note however, that it is often impossible to foresee results that might require the investigator to go back and repeat a test. A simple way reduce the occurrence of this is to go into a case well informed on how similar systems have failed. An excellent source 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 the expert and should be consulted regularly. Another important source of information are the standards by which the part was manufactured. These standards give the investigator a measuring stick by which to compare, as well as indicating areas of importance. There are many organisations that produce standards for different applications and several organisations standards can overlap. It would be a good idea for the investigators to spend some time familiarising themselves with these organisations and how the standards are used. Table P.1 gives a brief list of the more common organisations that write standards and their general area of coverage.

The first step in conducting any failure analysis is to gain a good understanding of the conditions under which the part was operating. The investigator must ask questions from those who work with, as well as those who maintain the equipment and visit the site whenever possible. Contacting the manufacturer may 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 will receive a failed part with little information about its history and operating conditions. In cases such as these the physical evidence will have to be more heavily relied on.

Figure P.1. Chart outlining the major steps that are usually taken when conducting a failure analysis.

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

The second step is to conduct a visual examination, cataloguing and recording the physical evidence at the 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 any area of particular importance, such as fracture surfaces and surface defects. Visual examination can be aided by the use of a stereomicroscope with lights that can be easily directed. Shadows can give depth to a surface making it easier to analysis and photograph. Pieces should always be examined and recorded before any surface cleaning is undertaken. In some cases substances such as dirt, paint and Oil on the surface can themselves be important clues, indicating such things as how old the fracture surface is and in what kind of environment the piece was operating. A good general rule is to be conservative when 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 characteristics that can be easily seen with the naked eye or the use of a stereomicroscope, however, sometimes a scanning electron microscope (SEM) will have to be used. There are several good books, some listed in the bibliography, on fracture mechanism and compilations of fracture surface photographs that can be used by the investigator to identify the mechanism of fracture under investigation. As a reminder, some common 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 background information the investigator must outline a plan of action, which is the series of steps that will be needed to successfully complete the case. There are several resources that an investigator can draw on to determine 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 be saved with a little forethought. The macroscopic examination is best performed when cataloguing the samples, however the investigator will often want to return to examine the part in more detail once other evidence is gathered. Use of a scanning electron microscope (SEM) is often useful at this stage because of its large range of magnifications and its large depth of field. Since undamaged fracture surfaces are not always available, it is often a good idea to open other cracks that may be present in the piece. This often reveals good quality fracture surfaces similar to those that caused failure. Procedures for doing this can also be found in the ASM handbook volume 10.

Nondestructive tests (NDT) are a good way to examine parts without causing permanent damage. Often times, 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 are available to the investigator and it would be a good idea to read up on each ones abilities. Table P.3 gives 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. Depending on 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 often depends on accessibility and sample type. In many cases, the SEM can be a powerful tool for fast identification of surface materials. Care should be taken not to contaminate samples taken for chemical analysis 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 the microstructure can be seen. Sometimes it is useful to take a cross section through the fracture surface so that the microstructure below the fracture and the surface profile can be examined. A section running parallel 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. There are 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 standards. There are many types of mechanical testing that can be performed and their procedures can be found in the ASTM mechanical testing standards. The most common method used is hardness testing because of its relative simplicity, low cost, and the fact that for many materials tables exist to relate hardness with yield strength. A macrohardness is usually sufficient to determine material properties, however microhardnesss measurements are helpful in determining property variations within the material. Use the microhardness measurement to compare the surface hardness to that of the body or to verify 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. This step can be difficult because when conducting the investigation clues will lead the investigator down paths 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 cases will 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 these reports is to demonstrate by example. Most of the cases mention the techniques that where used when stating the results. They where written at a basic level due to the uncertainty of background of the reader and further reading is be recommended to better understand the failure mechanism. Most of the cases that 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 downtime on an essential machine to production. The bolts were in operation on a crane used to transfer anodes into the machine. Figure 1.1 shows a drawing of the set-up and the location of fraction Just above the nut. 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 that conforms to SAE standards grade 5. The supplier of the new bolt confirmed that it was made to conform with ASTM standard A 193 grade B7.

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.

Results:

Observations

Examination of the fracture surface revealed characteristics such as a beachmarks associated with fatigue (Figure 1.3). The zone of final fracture was located between two areas of fatigue propagation suggesting the presence of bending forces. The surface area of final fracture was approximately 12% of the total fracture surface suggesting that the bolt was not overloaded. Cracks where also found between threads 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 content slightly below those required by the SAE standards for a grade 5 bolt. This lower carbon content would have acted to decrease the material properties. The chemical composition of the new sample bolt conformed to the ASTM standard A193/A grade B7 that requires an AISI-SAE 4140 composition.

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

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 before etching of the two bolts showed no cracking or unusually large inclusions. The original broken bolt did show 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 (Figure 1.5). The SAE grade 5 standard requires that the bolt be quenched and tempered to conform and therefore should have a tempered martensite structure. Martensite has higher material properties such as yield strength and hardness, which increases its resistance to fatigue initiation. The ferrite matrix of the original bolt has low yield strength, which in turn reduces its resistance to fatigue initiation. The new bolt was found to be quenched and tempered as required by the ASTM standard (Figure 1.6). However rolling seems where found at the tips of the treads (Figure 1.7). This is not a serious defect because of the defects location in a low stress area however, if the bolt was placed in a corrosive atmosphere these seams would corrode and then act as fatigue initiation sites.

Figure 1.4. Micrograph of flaking found at the base of a thread in the fractured bolt. 2% nital 100X	Figure 1.5. Micrograph of fractured bolt. Ferrite matrix with pearlite. 2% nital 200X	Figure 1.6. Micrograph of new bolt. Tempered martensite. 2% nital 500X	Figure 1.7. Micrograph of the new bolt thread showing a rolling seam. 2% nital 200X

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 are only two thirds that required by the standards. This conforms to the microstructural observations. The properties of the new bolt conformed to the standard even though they were slightly elevated.

Table 1.2--Results and standard requirements of tensile tests.
	Original Broken Bolt				New Sample Bolt		Standard 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.

Conclusions and Recommendations:

Examination revealed that the bolt failed as a result of high cycle low load fatigue. Chemical analysis and tensile tests confirmed that the bolt did not meet the SAE grade 5 standards required by the original design of the crane. The major cause for this lack of conformity is because the bolt was not quenched and tempered. Since the resistance of steel to fatigue initiation in proportional to its yield strength, the low 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 B7 bolt, as the supplier specified. However, rolling seams were found in the thread tips. Due to the relatively low loads this area is subjected to this is not a major problem but if the bolt is subjected to a corrosive 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 the upgrade 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 designed to ride on top of cardboard as it is being rolled. It was first installed in December 97 replacing a shaft in which 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 the shaft to become out of round by 0. 140". To repair this a hydraulic Jack was used at the centre of the roll to bend it back leaving a 0.040" deflection that was corrected by machining. Nine days later, on April 11th 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 installation to protect against corrosion in the mill environment. 17-4PH steel was used for this application before and failed to endure the high cycle low stress conditions.

Figure 2.1. Photograph of "rider roller" indicating approximate point of fracture.	Figure 2.2. Photograph of fracture surface showing initiation site, beachmarks from fracture propagation, and small area of final fracture.	Figure 2.3. Photograph of shaft surface indicating weld overlay flaw.

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 the material was adequate for the low applied stresses. The beachmarks (Figure 2.2), characteristics of fatigue that radiate from the initiation site, and the location of final fracture, being off centre, indicated that initiation did not occur evenly around the circumference of the shaft. Around the circumference of the fracture surface, a layer was observed which fractured at a 45' angle to the plane of fracture. This is characteristic of the weld overlay. As well, there were many grooves running around the outside of the shaft 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 low alloy 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 characteristics of low carbon steel (Figure 2.4). The weld overlay had pearlite matrix with some acicular ferrite (Figure 2.5). A microhardness test revealed a hard surface that gets progressively softer towards the core (Table 2.2). This concurs with the microstructure. The inclusions present in the core of the shaft where acceptable (Figure 2.6).

Examination of a longitudinal mount taken from near the point of major crack initiation sites showed large 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 shape and size to the inclusions. This suggests that these inclusions acted as stress raisers and thus as fatigue initiation sites. The bending of the shaft would have caused decohesion of the inclusions and increased the chances of fatigue initiation. Decohesion of the weld overlay between welding passes can also be seen around the circumference of the shaft (Figure 2.3). This indicates poor bonding between the weld overlay and the base material.

Figure 2.4. Micrograph of core microstructure composed of ferrite and pearlite. 2% nital 100X	Figure 2.5. Micrograph of weld overlay microstructure composed of a pearlite matrix with the presence of acicular ferrite. 2% nital 500X	Figure 2.6. Micrograph representing average inclusion content of the low carbon steel core. 2% nital 100X	Figure 2.7. Micrograph showing two inclusions found in the weld overlay 2% nital 15X	Figure 2.8. Micrograph showing the fracture surface initiation site. 2% nital 15X

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 conditions become an important factor in the prevention of crack initiation. Bending the shaft to correct its alignment probably caused decohesion of the weld inclusions encouraging microcracks to form. This would have increased local stress concentrations and the possibility of crack initiation. These inclusions probably 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 is applied property. This would harden the surface and thereby make the shaft more resistant to fatigue initiation at surface defects. A welding procedure should be developed that would involve the making of block samples in which the welding conditions, such as current and speed, are varied and optimised. Noranda Technology Centre can help in developing a procedure. A liquid penetrant inspection should be performed 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 35 HRC. The properties of this material fall between those of 1019 and 174PH. It will resist crack initiation better than the former, due to its higher endurance limit, and will resist crack propagation better than the latter, due to its higher fracture toughness (Table 2.3).

Table 2.3--Fatigue related properties of selected materials.
Material	Endurance limit (MPa)				Fracture Toughness
1018	275				260
4340	450				110
17-4PH	--				53

Other recommendations are:

• Avoid bending of shafts that have been surface hardened or had weld overlay applied due to the high possibility of inducing surface cracks.
• Avoid mechanical damage to the surface, such as scratches and dents, because they can act as crack 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 case study I was operating.

Figure 3.1a. Pin industrial drawing.	Figure 3.1b. Photograph of broken pin.

Observations :

Figure 3.2. Industrial drawing of pin chain and block mechanism.

The pin was broken in two locations approximately 2.4 and 5.2 centimetres from one edge. These locations are shown in relation to the mechanism in Figure 3.2. Examination of the surface revealed that where the bolt came in contact with the chain, sever plastic deformation was present. Examination of the 2.4 cm. fracture surface (Figure 3.3), which was located in an area of chain contact plastic deformation, revealed characteristics of fatigue. The fracture surface had little to no zone of final fracture indicating that the loads perpendicular to the fracture plane where low. Fatigue characteristics showed that fracture initiated on the opposite side to the deformation. This indicate that bending forces were present in the pin. Bending would have caused one side of the pin to be in tension and the other in compression. The fatigue 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 the adjoining 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 3.3. Photograph of fracture surface of 2.4 cm fracture.	Figure 3.4. Photograph of fracture surface of 5.4 cm fracture.	Figure 3.5. Photograph of pin indicating locations of fracture initiation.

A chemical analysis performed on the body of the pin revealed it to conform to the SAE AISI standard 1095. The original drawings for this application specify a SAE-AlSl 4140 (Table 3.1) Metallurgical examination of the mounted sample revealed plastic deformation at the edges as well as no significant inclusions. 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 fatigue propagation.

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 0.040 Sulphur 0.008 0.050 Chrome 0.03 -- Nickel 0.03 -- Molybdenum 0.01 --

Figure 3.6. Microphotograph of pin microstructure. Ferrite matrix with spherodised carbides. 2% nital 1000X

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.

Conclusions and Recommendations:

As the crane charges and unloads, the pin is subjected to bending forces. These forces create tensile forces 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 to fatigue initiation. For these reasons, the original design material, SAE AISI 4140 hardened to a range of 45 - 50 HRC, was a good choice.
• The block and chain should be examined for wear. If worn they would allow for larger bending then 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 in the 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 located on 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.

Figure 4.1. Photograph of bearing setup	Figure 4.2. Photograph of inner ring showing spalling in groove.	Figure 4.3. SEM photograph of spalling, flaking and cracking, in the groove. 200X	Figure 4.4. SEM photograph showing presence of 45º sheer planes. 500X

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 and cracking of the surface, was observed in the groove but was not evenly distributed around its circumference. Examination of the spalling using a scanning electron microscope (SEM) exposed flaking and the presence of surface cracks (Figure 4.3). Increased magnification of this area revealed fracture 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 is most severe. Beachmarks and river lines, which are characteristic of fatigue failures, revealed several initiation 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 lesser degree on the balls surfaces. The outer raceway revealed no major defects.

Figure 4.5. Photograph of the inner ring fracture surface.		Figure 4.6. SEM photograph of the inner ring fracture surface showing fatigue initiating at spall in the groove. 200X

Material characterisation and evaluation:

Both the compositions of the ball bearing and the inner raceway were found to fall within the norms for 52100 steel, AISI-SAE standards (Table 4.1). The microhardness measurements of both pieces are typical for this type of steel (Table 4.2). Surface hardness measurements for both ball and inner ring are similar, which is required by this type of application.

Microscopic examination of a cross section of the inner raceway revealed surface cracks consistent with the spalling observed (Figure 4-7). Etching the sample revealed a homogeneous macrostructure of a tempered martensite matrix with undissolved carbides present (Figure 4.8). This microstructure agrees with the chemical analysis and microhardness measurements.

Figure 4.7. Micrograph of cracks on the inner ring surface. 200X	Figure 4.8. Microphotograph of the inner ring microstructure composed of martensite and undissolved carbides. 2% nital 200X	Figure 4.9. Micrograph of cracks on the ball surface. 100X	Figure 4.10. Microphotographs of crack in a ball. 15X	Figure 4.11. Microphotograph of figure 4.10 etched with 2% nital showing heterogeneous martensite structure with undissolved carbides. 15X

Microscopic examination of a quartered ball bearing also revealed surface cracks (Figure 4.9). A large crack extending towards the centre of the bearing was also found (Figure 4.10). The microstructure is heterogeneous, unevenly distributed; tempered martinsite with undissolved carbides. The large surface crack ties along a border of the heterogeneity (Figure 4.11). Some decarburization was observed on the surface near spalling cracks.

Conclusions:

The failure was a result of vibrational fatigue initiated at spalling on the surface of the inner raceway. The spalling, which is a characteristic of contact fatigue, originated from the bearing being Installed Incorrectly or from it undergoing abnormal equiaxial radial loads in service, which caused a displacement of the inner ring. This displacement increased the axial loads causing the plastic deformation and spalling. Decarburization and uneven tempering of the balls as well as the extent of plastic 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 washer number 65B at a rate between 4 and 5 RPM. The contacting gear was a hardened steel worm gear, which was powered by a 50 horsepower 1800 RPM electric motor. The gear is a cast copper alloy with cut teeth and machined surface and was only in service for one month.

Figure 5.1. Photograph showing the bronze bull gear.		Figure 5.2. Photograph of the bull gear profile showing debris and severe materials loss.

Observations:

Examination of the gear tooth revealed that there was a large amount of material loss. A measurement taken near the base of the tooth where the material loss was most obvious revealed that tooth had gone from a thickness of 31 mm to 20 mm, a loss of I I mm. The contact surface had grooves running along the path the worm gear would have taken. Debris was also found along what was probably the exiting edge of the gear teeth (Figure 5.2). Along the front of the teeth, plastic deformation was seen near the edges where decreasing thickness could no longer support the load. Some cracking was observed in these areas. When opened, they revealed that the mode of crack propagation was interdendritic.

Chemical analysis of the bronze gear revealed that it conformed most closely with the UNS standard for copper alloy C90700 (Table 5.1). The lead and zinc content however were slightly above those allowed by the standard. Several samples where taken from the gear and examined microscopically. They revealed 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 in the material. The heterogeneity of the structure is caused by rejection of tin into solution as the dendrites grow while cooling. This segregation also reduces the mechanical properties of the material. Etching the microstructure with 20 nil NH40H, 20 ml H20, 20 nil H202 (3%) revealed a coarse dendrite microstructure (Figure 5.5). No plastic deformation of the working surface was observed which indicates abrasive wear.

Figure 5.3. Microphotograph showing the large amounts of porosity. 15X	Figure 5.4. Microphotograph showing interdendritic segregation. 200X	Figure 5.5. Microphotograph showing the large dendritic structure. 15X

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). This is below the Brinell-500 Kg hardness number of 95 (100VHN) required by the ASTM standard, B427-93a "Standard Specification for Gear Bronze Alloy Castings". A lower hardness number also suggests that the mechanical properties of the material would be below standards. This agrees with our metallographic examination.

Conclusions:

The bronze bull gear failed as a result of sever abrasive wear. The gear did not meet ASTM materials specifications for this application and this probably had a great influence on the final failure. However, there are several 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 hardness would 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 segregation combined to reduce the properties of the bronze bull gear below those required by ASTM B427-93a standards. A possibility is that that when the gear is subjected to loads or overloading, these low properties would allowed the gear teeth to deflect. The gear surfaces would no longer meet as they were designed, 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 B427-93a for the copper alloy UNS C 90700. As well the lubricant should be checked regularly for contamination and 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 service for 13 months when a leak was noticed. The reducer was installed on #1 acid storage tank, equipment number 50-200. The anodically protected carbon steel tank, contained off specification concentrated 93% sulphuric acid. The flow rate through the reducer was 400 gal/min.

Figure 6.1. Photograph of reducer.	Figure 6.2. (a-left) Old tank installation. (b-right) Tank installation at the time of reducer failure.

The tank was originally designed with a 4" diameter carbon steel nozzle, at floor level, that connected directly to a valve (Figure 6.2a). This lasted seven to eight years without incident. The design was changed to accommodate renovations so that an 8" carbon steel nozzle was installed 6" above the tank floor. This nozzle lead into the failed reducer, which then connected to a valve composed of alloy 1-0 steel (Figure 6.2b). This valve was said to be badly corroded. The valve then led to a 6" pipe made of 316L stainless steel in which no problems were found. After the reducer failure, the piping arrangements were 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 (Figure 6.3). This area, which was probably an air pocket, extended from the top of the 87' diameter flange into the reducing pipe where is stopped just before the 6" diameter flange. Damage in this area consisted of minor pitting (Figure 6.4). Damage, resembling a honeycomb structure in places, was most severe just below the air pocket in the reducing pipe near the 6" diameter end (Figures 6.5a and 6.5b). This is where the leak was found (Figure 6.6). The damage becomes less severe in the pipe section towards the bottom. Only pitting was found in both the 8" and 6" flanges.

Figure 6.3. Photograph of the top insider of the reducer showing the area at the top where little damage occurred.	Figure 6.4. Microphotograph of pitting in air pocket. 15X	Figure 6.5. Photographs showing areas to the (a-left) right and (bright) left of the top relatively undamaged surface. The red arrow in (a) indicated where the leak occurred.

Chemical analysis of the flange and the pipe revealed that they both conform to AISI-SAE standards for 316L stainless steel (Table 6.1)

Closer examination of the inside surface of the reducer with a SEM revealed dimples (Figure 6.7). These features are typical of a ductile deformation, which indicates abrasion. The orientation of the features also follows the direction of liquid flow. Pitting and uniform corrosion was also found in the region (Figure 6.8).

Figure 6.6. Photograph taken on the outside of the reducer showing the hole where the reducer leaked.	Figure 6.7. SEM photograph of the inside surface of the reducer in the damaged area. 200X	Figure 6.8. SEM photograph of the inside surface of the reducer in the damaged area. 500X

Conclusions and Recommendations:

A combination of two mechanisms caused the failure. Severe turbulence in the reducer caused a degradation of the passive layer that protects the stainless steel from corrosion. This would have left the system open to severe corrosion, which in turn would have lead to failure. The top of the reducer was probably protected by the presence of an air pocket.

The second mechanism was erosion, originating when air bubbles near the surface imploded causing mechanical damage, cavitation. Turbulence in the system may have formed bubbles from the air pocket at the top of the reducer. These bubbles would then have been carried into the reducer where increasing pressures would have caused them to implode. The highly corrosive environment would have increased the rate of degradation dramatically.

The new setup, placing the valve before the reducer, changed the dynamics of the system and may have solved the problem, however existing reducers and valves should have their thickness monitored at regular intervals using an ultrasonic thickness gauge. If problems reoccur, the system should be evaluated for excessive turbulence and air pockets. A possible solution would be to use a PTFE liner in the reducer. This would provide a barrier that protects against turbulence but not cavitation.

APPENDIX 1: EXAMPLE QUESTIONNAIRE

Bibliography