Fatigue failure of engineering components and structures results from progressive fracture caused by cyclic or fluctuating loads. Fatigue is an important potential cause of mechanical failure, because most engineering components or structures are or can be subjected to cyclic loads during their lifetime. This article focuses on fractography of fatigue. It provides an abbreviated summary of fatigue processes and mechanisms: fatigue crack initiation, fatigue crack propagation, and final fracture,. Characteristic fatigue fracture features that can be discerned visually or under low magnification are then described. Typical microscopic features observed on structural metals are presented subsequently, followed by a brief discussion on fatigue in polymers and polymer-matrix composites.

Fatigue failures may occur in components subjected to fluctuating (time-dependent) loading as a result of progressive localized permanent damage described by the stages of crack initiation, cyclic crack propagation, and subsequent final fracture after a given number of load fluctuations. This article begins with an overview of fatigue properties and design life. This is followed by a description of the two approaches to fatigue, namely infinite-life criterion and finite-life criterion, along with information on damage tolerance criterion. The article then discusses the characteristics of fatigue fractures followed by a discussion on the effects of loading and stress distribution, and material condition on the microstructure of the material. In addition, general prevention and characteristics of corrosion fatigue, contact fatigue, and thermal fatigue are also presented.

Thermomechanical fatigue (TMF) is the general term given to the material damage accumulation process that occurs with simultaneous changes in temperature and mechanical loading. TMF may couple cyclic inelastic deformation accumulation, temperature-assisted diffusion within the material, temperature-assisted grain-boundary evolution, and temperature-driven surface oxidation, among other things. This article discusses some of the major aspects and challenges of dealing with TMF life prediction. It describes the damage mechanisms of TMF and covers various experimental techniques to promote TMF damage mechanisms and elucidate mechanism coupling interactions. In addition, life modeling in TMF conditions and a practical application of TMF life prediction are presented.

A structure had been undergoing fatigue testing for several months when a post-like member heat treated to a tensile strength of 1517 to 1655 MPa (220 to 240 ksi) ruptured. The fracture occurred in the fillet of the post that contacted the edge of a carry-through box bolted to the member. At failure, the part was receiving a second set of loads up to 103.6% of design load. Visual investigations showed rubbing and galling of the fillet. Microscopic and metallographic examination revealed beach marks on the fracture surface and evidence of cold work and secondary cracking in the rubbed and galled area. Electron fractography confirmed that cracking had initiated at a region of tearing and that the cracks had propagated by fatigue. Mechanical properties of all specimens exceeded the minimum values specified for the post. This evidence supports the conclusion that fatigue was the primary cause of failure. Rubbing of the faying surfaces worked the interference area on the post until small tears developed. These small tears became stress-concentration points that nucleated fatigue cracks. Recommendations included rounding the edge of the box in the area of contact with the post to ensure a tangency fit.

A pushrod made by inertia welding two rough bored pieces of bar stock installed in a mud pump fractured after two weeks in service. The flange portion was made of 94B17 steel, and the shaft was made of 8620 steel. It was disclosed by visual examination that the fracture occurred in the shaft portion at the intersection of a 1.3 cm thick wall and a tapered surface at the bottom of the hole. The fatigue crack was influenced by one-way bending stresses initiated at the inner surface and progressed around the entire inner circumference. A heavily decarburized layer was detected on the inner surface of the flange portion and sharp corner was found at the intersection of the sidewall and bottom of the hole. It was concluded that the stress raiser due to the abrupt section change was accentuated by decarburized layer. As a corrective measure, the design of the pushrod was changed to a one-piece forging and circulation of atmosphere during heat treatment was permitted through a hole drilled in the flange end of the rod to avoid decarburization.

Plunger shafts machined from 4150 steel bar stock were involved in a series of fatigue failures. The fractures consistently occurred at two locations on the shafts: the shaft fillet and either side of a machined notch. The material specification for the shafts required 41xx series steel with a carbon content of 0.38 to 0.53%, a hardness of 35 to 40 HRC for the shaft, and a hardness of 50 to 55 HRC for the notch (which was case hardened). Analysis (visual inspection, chemical analysis, hardness testing, and magnetic particle inspection) supported the conclusions that all the fractures were fatigue-induced failures due to sharp radii in the fillets. The stress-concentrating effects of the fillets caused fatigue cracks to initiate and grow under cyclic loading until the crack depth was critical, causing the shaft to fail and rendering the assembly inoperative. Recommendations included increasing the radii of the notch and shaft fillets. If fatigue cracking had continued to be a problem with this component, shot peening of the subject radii would be appropriate. This process produces residual compressive stresses in the surface of the part, thereby retarding initiation of fatigue cracks.

Stainless steel liners (AISI type 321) used in bellows-type expansion joints in a duct assembly installed in a low-pressure nitrogen gas system failed in service. The duct assembly consisted of two expansion joints connected by a 32 cm (12 in.) OD pipe of ASTM A106 grade B steel. Elbows made of ASTM A234 grade B steel were attached to each end of the assembly, 180 deg apart. A 1.3 mm (0.050 in.) thick liner with an OD of 29 cm (11 in.) was welded inside each joint. The upstream ends were stable, but the downstream ends of the liners remained free, allowing the components to move with the expansion and contraction of the bellows. Investigation (visual inspection, hardness testing, and 30x fractographs) supported the conclusion that the liners failed in fatigue initiated at the intersection of the longitudinal weld forming the liner and the circumferential weld by which it attached to the bellows assembly. Recommendations included increasing the thickness of the liners from 1.3 to 1.9 mm (0.050 to 0.075 in.) in order to damp some of the stress-producing vibrations.

The case and stiffener of an inner-combustion-chamber case assembly failed by completely fracturing circumferentially around the edge of a groove arc weld joining the case and stiffener to the flange. The assembly consisted of a cylindrical stiffener inserted into a cylindrical case that were both welded to a flange. The case, stiffener, flange, and weld deposit were all of nickel-base alloy 718. It was observed that a manual arc weld repair had been made along almost the entire circumference of the original weld. Investigation (visual inspection, 0.5x macrographs, and 10x etched with 2% chromic acid plus HCl views) supported the conclusions that failure was by fatigue from multiple origins caused by welding defects. Ultimate failure was by tensile overload of the sections partly separated by the fatigue cracks. Recommendations included correct fit-up of the case, stiffener, and flange and more skillful welding techniques to avoid undercutting and unfused interfaces.

After being in service for ten years the ball-and-race coal pulverizer was investigated after noises were noted in it. Its lower grinding ring was attached to the 6150 normalized steel outer main shaft while the upper grinding ring was suspended by springs from a spider attached to the shaft. A circumferential crack in the main shaft at an abrupt change in shaft diam just below the upper radial bearing was revealed by visual examination. The smaller end of the shaft was found to be slightly eccentric with the remainder when the shaft was set up in a lathe to machine out the crack for repair welding. The crack was opened by striking the small end of the shaft and the shaft was broken 1.3 cm away from the crack in the process. A previous fracture that resulted from torsional loading acting along a plane of maximum shear was revealed almost perpendicular to the axis of the shaft. Faint lines parallel to the visible crack thought to be fatigue cracks were revealed on examination of the machined surface. The shaft was repaired by welding a new section and machined to required diameters and tapers to avoid abrupt changes.

A stepped drive axle (hardened and tempered resulfurized 4150 steel forging) used in a high-speed electric overhead crane (rated at 6800 kg, or 7 tons, and handling about 220 lifts/day with each lift averaging 3625 to 5440 kg, or 4 to 6 tons) broke after 15 months of service. Visual examination of the fracture surface revealed three fracture regions. The primary fracture occurred approximately 50 mm (2 in.) from the driven end of the large-diam keywayed section on the stepped axle and approximately 38 mm (1 in.) from one end of the keyway where the crane wheel was keyed to the axle. Macroscopic, microscopic, and chemical examination revealed composition that was basically within the normal range for 4150 steel. This evidence supports the conclusion that cracking initiated at a location approximately opposite the keyway, and final fracture was due to mixed ductile and brittle fracture. Axial shift of the crane wheel during operation, because of insufficient interference fit, was the major cause of fatigue cracking. Recommendations included redesigning the axle to increase the critical diameter from 140 to 150 mm (5.5 to 6 in.) and to add a narrow shoulder to keep the drive wheel from shifting during operation.

The crane hook (rated for 13000 kg) failed in the threaded shank while lifting a load of 9072 kg. The metal in the hook was revealed by chemical analysis to be killed 1020 steel. It was disclosed by visual examination that the fracture had at the last thread on the shank and rough machining and chatter marks were evident on the threads. Beach marks that emanated from the thread-root locations on opposite sides of the fracture surface identified these locations to be the origins of the fracture. A medium-coarse slightly acicular structure was revealed by metallographic examination which indicated that the material was in the as-forged condition (which meant lower fatigue strength). The fracture was concluded to have occurred due to stress concentration in the root of the last thread. Normalizing of the crane hook after forging was suggested as a corrective measure. A stress-relief groove with a diam slightly smaller than the root diam was placed at the end of the thread and a large-radius fillet was machined at the change in diameter of the shank.

A 10,890-kg coil hook torch cut from 1040 steel plate failed while lifting a load of 13,600 kg after eight years of service. The normal ironing (wear) marks were exhibited by the inner surface of the hook. It was revealed by visual examination that cracking had originated at the inside radius of the hook. Beach marks (typical of fatigue fracture) were found extending over approximately 20% of the fracture surface. Numerous cracks were revealed by macroscopic examination of the torch-cut surfaces. It was revealed by macrograph of an etched specimen that the cracks had initiated in a hardened martensitic zone at the torch-cut surface and had extended up to the coarse pearlite structure beneath the martensitic zone. The fatigue fracture was concluded to have initiated in the brittle martensitic surface while failure was contributed by the 25% overload. As a corrective measure, the coil hooks were flame cut from ASTM A242 fine-grain steel plate, ground to remove the material damaged by flame cutting and stress relieved at 620 deg C.

A bridge wheel from a 272,160 kg stripper crane fractured in the web near the rim after one year of service. The wheel was forged from 1055 steel, and the tread, hub faces, and hub bore were machined. Beach marks indicative of fatigue at ten locations were revealed by macroscopic examination of the fracture surfaces. The surface of the web was heavily scaled and decarburized. A gross forging defect extending about 1.8 mm along the fracture surface was disclosed by examination of a micrograph of a section through one of the fatigue origins. Shallower forging defects were visible along the web surface. Fatigue cracking of the wheel was initiated at forging defects in the web. Replacement wheels were machined all over and were magnetic particle inspected to detect any cracks that could act as stress raisers.

A drum pinion shaft (1030 steel) which was part of the hoisting gear of a crane (capacity 18,140-kg) operating in a blooming mill failed while lifting a 9070 kg load. Chatter marks, rough-machining marks, and sharp corner radii were revealed in the keyway which extended into a shoulder at a change in diam. A circular recess below the keyway surface was revealed at each end of the keyway. A sharp corner at the end of the keyway was revealed by examination to be the origin of fracture. Beach marks were found radiating from the origin over a large portion of the fracture surface which confirmed failure of the shaft by fatigue fracture. As a corrective measure the shaft was replaced with one made of 4140 steel, quenched and tempered to a hardness of 286 to 319 HRB. The keyway was moved away from the change in section and was machined with a 1.6-mm radius in the bottom corners and a larger-radius fillet was machined at the change in section.

The 16 mm diam 6 x 37 fiber-core improved plow steel wire rope on a scrapyard crane failed after two weeks of service under normal loading conditions. This type of rope was made of 0.71 to 0.75% carbon steel wires and a tensile strength of 1696 to 1917 MPa. The rope broke when it was attached to a chain for pulling jammed scrap from the baler. The rope was heavily abraded and several of the individual wires were broken. a uniform cold-drawn microstructure, with patches of untempered martensite in regions of severe abrasion and crown wear was revealed by metallographic examination. As a result of abrasion, a hard layer of martensite was formed on the wire. The wire was made susceptible to fatigue cracking, while bending around the sheave, by this brittle surface layer. The carbon content and tensile strength of the wire was found lower than specifications. As a corrective measure, this wire rope was substituted by the more abrasion resistant 6 x 19 rope.

A stuffing box (sand cast from ASTM A 536, grade 60-45-10, ductile iron) began leaking water after two weeks of service. The machine was operating at 326 rpm with a discharge water pressure of 21.4 MPa (3100 psi). Investigation (visual inspection, mechanical analysis, and nital etched 100x magnification) supported the conclusion that the crack initiated at the inner edge of a lubrication hole and had propagated toward both the threaded and flange ends of the casting. An appreciable residual-stress concentration must have been present and caused propagation of the crack. The residual stress might have been caused when a fitting was tightly screwed into the lubrication hole, and it might have been concentrated by notches at the inner end of the hole created when the drill broke through the sidewall to the stuffing box.

A bolt breaks along a change in cross section well below its rated capacity. An anchoring screw spins freely in place, having snapped at its first supporting thread. A motor unexpectedly disengages its load, its driveshaft having fractured near a keyway. Such failures – involving axles, leaf springs, engine rods, wing struts, bearings, gears, and more – can occur, seemingly without cause, due to vibrational fracture. Vibrational fractures begin as cracks that form under cyclic loading at nominal stresses which may be considerably lower than the yield point of the material. The fracture is proceeded by local gliding and the development of cracks along lattice planes favorably orientated with respect to the principal stress. This non-reversible process is often misleadingly called “fatigue” and presents significant challenges to engineering teams that ill-advisedly take to searching for material faults. Several examples of notch-induced vibrational fractures are presented along with guidelines for investigating their cause.

The fan drive support shaft, specified to be made of cold-drawn 1040 to 1045 steel, fractured after 2240 miles of service. It was revealed by visual examination of the shaft that the fracture had initiated near the fillet at an abrupt change in shaft diameter. The cracks originated at two locations approximately 180 deg apart on the outer surface of the shaft and propagated toward the center. Features typical of reversed-bending fatigue were exhibited by the fracture. A tensile specimen was machined from the center of the shaft and it indicated much lower yield strength (369 MPa) than specified. It was disclosed by metallographic examination that the microstructure was predominantly equiaxed ferrite and pearlite which indicated that the material was in either the hot-worked or normalized condition. An improvement of fatigue strength of the shaft by the development of a quenched-and-tempered microstructure was recommended.

A resistance-welded chain link made from 16 mm diam 4615 steel failed while lowering a 9070 kg load of billets into a rail car after being in service for 13 months. Beach marks, typical of fatigue were found to have originated at the inside of the link which broke at the weld. Cracks in the weld zone (up to 1.2 mm deep) were revealed during metallographic examination of a section through the fracture surface. The cracks were filled with scale which indicated that they had formed during resistance welding of the link. The defect was thus attributed to the weld defects which initiated the fatigue failure by acting as stress raisers. The welding method was changed by the manufacturer and all chains were replaced with defect free chains.

Several deburring drums that fractured were filled with abrasive, water, and small parts, such as roller bearing rollers, and rotated on their axis at 36 rpm. Cracks were discovered very early in the service lives of these high-chromium white iron cast structures. All of the fractures were through bolt holes in the mounting flange. The holes had a sharp edge and exhibited uneven wear on the inside diameter. In operation, the mounting bolts were frequently found to be loose and in at least one case broken off. A 25x scanning electron microscopy (SEM) fractograph from near this fracture-initiation area showed fatigue striations. No casting or metallurgical structural defects were found that could explain the failures. This evidence supports the conclusion that cracking was a result of the stress-concentration site at the bolt holes where a fatigue-initiated fracture occurred. Recommendations included that the radii be increased at the sharp corners and that lock-wiring be used to secure against bolt loosening.