Fastener Fatigue

What is Fatigue?
 

Fatigue is an insidious failure mode for a material to suffer from because the crack usually grows slowly until an abrupt catastrophic failure occurs. An application may be in operation for days, weeks, months, or years without any signs of problems before, suddenly, fasteners begin to break. ASTM E1823-10a, Terminology Relating to Fatigue and Fracture Testing, defines fatigue as:

The process of progressive localized permanent structural change occurring in a material subjected to conditions that produce fluctuating stresses and strains at some point or points and that may culminate in cracks or complete fracture after a sufficient number of fluctuations.

A key quote in the above definition is “...a material subjected to conditions that produce fluctuating stresses and strains....” To give an example of a fatigue application, picture a wheel stud. Wheel studs are subject to vibrations and shock loads from the irregularities in a road’s surface; our wheels, at times, take a beating. This cyclical loading is transferred to our auto’s suspension via the wheel studs. Our saving grace is that wheel studs are wonderfully overdesigned with a hefty safety factor to ensure a fatigue failure has an extremely low probability of occurring.

Any application that produces an oscillating/fluctuating/cyclical stress or strain can cause a fatigue failure if not appropriately designed. Potential initiators of fatigue include vibration, thermal expansion and contraction, shock loading, and improper joint design.

 
What Does a Fatigue Failure Look Like?
 

Fatigue failures usually have flat profiles with very little topography and very little plastic deformation or necking. Many well-intentioned persons describing a fatigue failure for the first time will say the fastener “sheared,” as a shear failure is generally understood to be a flat fracture. Shear and fatigue, however, are very different failure modes and have quite different causes and solutions. The common look to a fatigue failure is shown below in Figure 1.

Figure 1 (Left) Figure 1 (Right)
Figure 1: Top down view (left) and profile view (right) of a failure where reversed bending fatigue propagated nearly 100% across the fracture surface.

Landmark features of fatigue that can usually be seen by the naked eye include ratchet lines, progression marks and sometimes river marks. Ratchet lines initiate from the surface and extend along with the crack propagation direction (Figure 2). Ratchet lines are most commonly perpendicular to the surface, however, because they grow along with the fatigue crack, they will extend in whichever direction the crack is developing.

Figure 2
Figure 2: Many ratchet lines are seen in the image above. Two ratchet lines in this Grade 8 hex cap screw (HCS) are highlighted by red arrows. Direction of crack propagation is along the Y-axis. 10x.

A ratchet line comes from two crack planes converging onto a single plane (Figure 3) which eventually leads to planes coalescing and the ratchet line disappears. Each plane adjacent to a ratchet line contains a separate crack origin. The more ratchet lines observed, the greater the applied stress and/or the greater the stress concentration was.

Figure 3 (Left) Figure 3 (Right)
Figure 3: Schematic view of a fatigue crack and single ratchet line that initiated on the exterior of the sample in a thread root. In the case above, tensile load is applied along the Z-axis and the crack is propagating along the Y-axis. Crack planes 1 and 2 initiated on different X-Y planes. As crack propagation continued, they coalesced to form crack plane 3 and thus the ratchet line disappears.

Progression marks are also known as arrest lines or “beach marks” because they take on the appearance of sand on a beach after waves have repeatedly washed over the shoreline. Notice in Figure 4 of a stainless steel socket head cap screw (SHCS) that as the crack progresses further away from the origin, the more spaced out the progression marks become. This is an indication stress is increasing as the crack propagates and as cross-sectional area is lost due to a growing crack. Note that it can be possible to see each incremental load cycle, called a fatigue striation, though it requires high magnification that optical microscopes cannot achieve (roughly 5,000x).

Figure 4
Figure 4: Arrest lines in an M20-2.5x60mm, A2-70, SHCS (304 stainless steel). Direction of crack propagation is along Y-axis. 50x.

Ratchet lines and progression marks come in all different appearances; like fingerprints, one fracture is never identical to another. Depending on the appearance of the fracture surface, you can generally categorize the type of loading the part experienced into (1) tension-tension/tension-compression, (2) unidirectional bending, (3) reversed bending, (4) rotational bending, and (5) torsion. See Figures 5 and 6.

Figure 5
Figure 5: Arrows indicate the direction of loading. Note, torsion fatigue is very uncommon in fasteners and therefore was not shown here.
Figure 6
Figure 6: Images of loading types with high nominal stress and low nominal stress. Mild or severe stress concentrations are the most typical comparisons for fasteners. Figure from G.F. Vander Voort, Visual Examination and Light Microscopy, ASM Handbook, Volume 12: Fractography, ASM International, 1987, p 118.

While Figure 6 shows the resultant appearance of known single condition loading, i.e. just unidirectional bending or just rotational bending, often fasteners experience multiple loading conditions where, e.g. you may see remnants of rotational bending that was followed by unidirectional bending. Figures 7-10 are a few real life examples of fatigue fractures that will provide a greater understanding of the complexity of the loading conditions fasteners experience.

Another landmark feature of fatigue sometimes observed is called river marks. River marks point in the direction of crack propagation. An example of river marks is shown in Figure 7 of a 2 1/2" diameter ASTM A354 Grade BD bolt. Progression marks cover 100% of the fracture surface indicating there was essentially no axial load applied to this fastener when final fracture occurred. The loading condition of this fastener was unidirectional bending fatigue. This fracture surface shows many ratchet lines along the thread root indicating the thread acted as a severe stress concentration.

Figure 7
Figure 7: This is a 2 1/2" Grade BD bolt. Direction of crack propagation is top to bottom. River marks (A) and ratchet lines (B) are observed. Disregard post fracture: spot corrosion (dark spots) and abrasion marks (appear lustrous/white).

Figure 8 is an M12 diameter ISO 898-1 Class 12.9 SHCS. Lighter regions (A) of this fracture are progression marks. The dark region (B) is the area of fast fracture (not fatigue), or the area that was still holding the tensile load just before final fracture. There was a significant percent cross sectional area remaining when fast fracture occurred, indicating the part had a significant clamp load when the part finally fractured.

Figure 8
Figure 8: M12 Class 12.9 SHCS. Note that region A contains signs of rotational bending and unidirectional bending fatigue.

Figure 9 shows two 5/8” diameter SAE J429 Grade 8 HCS that were taken from the same application, yet experienced significantly different types of loading. How it was possible to achieve such different loading conditions on each bolt is something that looking only at the fracture surface will not tell us. To understand how the loading conditions were different a great deal more information is required of the history of the application, e.g., joint materials, assembly instructions, maintenance programs, external loads.

Figure 9
Figure 9: 5/8” Grade 8 HCS. Nearly 100% of the fracture surface is covered by progression marks. Bolt 1 experienced rotational bending (A), unidirectional bending from left to right (B) and unidirectional bending from bottom to top (C). Bolt 2 experienced reversed bending (D) and slight rotation primarily unidirectional bending (E).

Figure 10 is a 1 1/8” diameter SAE J429 Grade 8 HCS. This sample also experienced a combination of various types of loading: unidirectional bending, reversed and slight rotational bending, and reversed bending. The darker region is the area of fast fracture which was the last area carrying load when the part finally broke.

Figure 10
Figure 10: 1 1/8” Grade 8 HCS. Approximately 85% of the part is covered by progression marks and 15% is fast fracture. This bolt experienced unidirectional bending (A), reversed and slight rotational bending (B, C), and finally fast fracture/overload (D). In region B the progression marks were closely spaced, characteristic of low load, high cycle fatigue. In region C the progression marks were more widely spaced, characteristic of high load, low cycle fatigue. Disregard post fracture: spot corrosion.
 
Why do Fasteners Fail in Fatigue?
 

Why fasteners fail in fatigue can generally be broken down into issues stemming from design, installation, and loosening over time (or combinations thereof). Note, this article does not consider defective fasteners that have failed in fatigue. The reason for this is the answer to the question of why a fastener with an abnormal stress concentration failed in fatigue is obvious: the defect! However, when a fastener conforms to all requirements of the fastener specification or print yet still fails is what this section of the article intends to address as the answer to why the part broke in these cases can be more elusive.

Design Issues:

  • Fasteners are essentially springs, and we stretch these springs to induce a preload on joint materials to rigidly hold the clamped members together. Some of these springs are soft and ductile, such as ASTM A307 Grade A, SAE J429 Grade 2, and ISO 898-1 Class 4.6. Other springs are strong and stiff, such as ASTM A574, Grade 9, and ISO 898-1 Class 12.9. Each spring is designed to be used with like materials, e.g.: ASTM A307 Grade A might be specified to bolt together a soft steel like ASTM A36 plate; ASTM A574 SHCS might be specified to bolt together a very strong steel such as tool steel. The reason an ASTM A574 SHCS is not recommended to be used with an ASTM A36 plate is because the A36 will plastically flow under the recommended clamp loads that an A574 fastener induces.
  • A fastener's stiffness can be adjusted through the material and strength selection process. Similarly, a joint’s stiffness can be adjusted by selecting material and strength but also adjusting geometry. Joint geometry is significant because a larger compression barrel area will increase joint stiffness. E.g., a short fastener (thinner joint members) will induce a small joint compression barrel while a long fastener (thicker joint members) will induce a large compression barrel. This compression barrel area is what we use to calculate the stiffness of this joint. Unfortunately, how to calculate compression barrels goes beyond the scope of this article; for additional information on this topic, please contact Fastenal engineering at engineer@fastenal.com, or review VDI 2230 BLATT 1, titled Systematic calculation of high duty bolted joints – Joints with one cylindrical bolt.
  • Bolt diagrams are useful in describing how changing the resilience/stiffness of the bolt and joint members will influence the cyclical load acting on the fastener. Read more about bolt diagrams in the Bolted Joint Design article. Figure 11 shows three different conditions of joint member resilience/stiffness. Note the load amplitude the fastener observes determines fatigue life: a fastener with a small load amplitude will have a long fatigue life, and a fastener with a large load amplitude will have a short fatigue life.
  • Figure 11 (Top)
    Figure 11 (Center)
    Figure 11 (Bottom)
    Figure 11: Bolt diagram for a stiff joint member (A), a moderately stiff joint member (B) and a non-stiff joint member (C). As joint stiffness goes to ∞ (as drawn above, slope would be -∞), FB goes to zero and the load acting on the bolt becomes static, thus no fatigue. As the joint members become less stiff (more resilient), FB increases and bolt fatigue life decreases.
    FI = initial preload in the fastener/joint from installation. This is sometime called the target clamp load.
    FE = external applied load while the application is in use.
    FB = observed cyclical load on the bolt.
    FJ = observed cyclical load on the joint members.
  • Even if a joint is properly designed where bolt resilience and joint stiffness have been considered, it is still possible for that joint to experience a fatigue failure. Most commonly these fatigue failures are caused by poor installation practices (e.g. not achieving target clamp load) or the fastener loosening over time while the application was in use (e.g. vibration loosening).

Installation Issues:

  • Insufficient Preload
    A lower than designed preload would mean the fastener will observe a higher cyclical load amplitude and thereby lower fatigue life. Different tightening strategies achieve different levels of target clamp load accuracy. One of the most common issues that leads to fatigue is under-tightening and is typically a consequence of torque control or lack thereof. When using predefined torque values from a chart, the resultant preload may be ±30% from the targeted preload value. This scatter is due to a multitude of variables with friction playing a significant role. A nut factor (also known as torque-coefficient or k-factor), is often used to condense the various traits and conditions that affect preload from a given torque, and this constant is often derived empirically due to the difficulty in its prediction. For further information discussing tensioning methods, torque control, nut factor and its variables read the Bolted Joint Design article.
  • Retightening Post Relaxation
    For joint members that are as hard as or harder than the fastener, there is always some level of joint relaxation post fastener installation. Relaxation usually completes within the first hour after installation, and usually the loss of load is on the order of 1-2% of the initial clamp load; as the hardness of the joint members decreases, the percent of clamp load lost increases. Note that if there is a significant disparity between the hardness of the fastener and the joint materials this could lead to the fastener embedding into the joint member which will be discussed later. If possible, a final torque pass should be completed to regain this lost clamp load.
  • A Lack of Fastener Lubrication
    When fasteners are installed via torque control adding a lubricant to the threads and under-head bearing surface significantly reduces scatter in clamp load. The proper preloading of fasteners is of greater influence on maintaining clamp load than is lubricity on the threads, which would induce loosening. If the joint is tight and no joint member slip is permitted, loosening will not occur. When preload control is critical, the use of lubrication is recommended.

Loosening Over Time:

  • Improper Fastener and Material Selection
    In designing a joint, the joint material stiffness should be significantly greater than the bolt stiffness. The joint should be a rigid body capable of supporting the full clamp load of the fastener without sacrificing any structural integrity. This includes selecting the right externally threaded fasteners to go with the internally threaded component.
    • Grade 5 HCS installed into Styrofoam tapped hole
      This is an over exaggeration of reality, but it demonstrates the point. You could never torque the Grade 5 to the appropriate torque to achieve the right preload a Grade 5 fastener needs to function properly. If you did try and torque to Grade 5 torque recommendations, the Styrofoam would simply plastically flow, strip out, and be crushed.
    • Class 12.9 SHCS installed into Class 8 nut
      The same rules apply to this scenario as the Grade 5 installed into Styrofoam; the Class 12.9 will never be able to achieve the right preload, and stretch, to function properly. The main problem is the torque required to properly seat the Class 12.9 would likely strip the Class 8 nut, and, even if the nut did not strip, the nut would have significant plastic flow and relaxation and therefore clamp load would be reduced.
  • Vibration / Dynamic Loading
    Vibration by its very definition is a cyclical loading and can cause loosening of a joint. As discussed, once the joint is loose, external loads are applied directly to the fastener, and fatigue is possible. Adding friction to the joint members’ bearing surface (reference Figure 12, surface between Plates I and II) will reduce the possibility of joint members slipping and thereby reduce the possibility of the fastener loosening.
  • Thermal Expansion and Contraction
    At times fasteners are used to join various types of metals and alloys, e.g. carbon steel, alloy steel, aluminum, brass/bronze, cast iron, etc. All these materials have slightly different coefficients of thermal expansion and when this conglomeration of materials is used together in a joint with fluctuating temperatures, the difference in expansion and contraction rates can cause undue stress and strain on the fastener.
  • Embedment
    If the mating joint material the fastener is in contact with is too soft to support the fastener’s required clamp load, embedment (plastic deformation of the joint material) will occur. Over time, dynamic loads acting on the joint can further embed the fastener and displace joint material. Joint relaxation and embedment in some cases go hand-in-hand, and you can get embedment during installation. This is another reason to make sure the bearing joint member is harder than the fastener being used.
 
What Ways Are There to Combat Fatigue?
 
  • Fastener Geometry
    Fasteners, which we think of as springs with shanks, limit the region over which the greatest stretching occurs, i.e. confining stretching to the threaded region within the grip range (see Figure 12A length X). This is because the smaller cross section area will incur more stretching per unit length. If the fastener stretches due to some dynamic loading or vibrations, ideally stretching will be uniform and distributed over a long length to minimize stress concentration. Going from a standard HCS to a tap bolt is one option to reduce the stress concentration. Because tap bolts are threaded their entire length, they have uniform cross sectional area from under head surface to first engaged thread. See Figure 12B, length X+Y. Another option is to significantly modify the geometry of the fastener as shown in Figure 12C. Any stretching that occurs is forced to act in the much longer turned down and polished region thus distributing stress concentrations and increasing fatigue life.
    Figure 12
    Figure 12: Standard HCS where uniform stretching is confined to length X (A). Tap bolt (B) and modified geometry (C) where the length of bolt for which uniform stretching occurs increases to X + Y.

    Increasing the grip range (length X+Y in Figure 12) and thereby increasing the length of the fastener will also improve fatigue life. This is accomplished by the clamp barrel area increasing which increases joint stiffness. Reference Figure 11.
  • Polishing
    Surface conditions like nicks, dents, and tool marks are stress concentrations that could act as crack initiation sites. By removing or reducing crack sizes via polishing, we remove the locations were a fatigue crack could initiate. Thus, fatigue life can be improved. It is important to remember mass production fasteners will naturally contain surface conditions as parts tumble along the production process.
  • Thread Considerations
    If the design engineer knows their application could experience fatigue loading, consideration should be given to how the threads are formed and the thread profile.
    • Cut threads can have rougher thread surfaces due to chatter or tool marks. Cutting interrupts grain flow, exposing grain boundaries, which can lead to crack initiation sites and corrosive attack sites. Rolled threads typically have smoother thread surfaces and generally greater thread strength due to work hardening. Rolling threads provides an unbroken grain flow of the material in the region of the threads. Rolled threads are the better option when considering a fatigue application. Note some fastener specifications require rolled threads for certain fastener diameter ranges and for other diameter ranges permit rolled or cut threads. The designer of the joint should be mindful what the fastener specifications permit for how threads are formed and determine if additional verbiage is required on a print to specify thread forming process. Threads are commonly formed before heat thread; rolled threads after heat treat is yet another option available to a designer who wishes to retain as much compressive force in the threads as possible.
    • Specifying thread form such as UNR, UNJ or MJ will improve fatigue performance. Unified national rounded (UNR) root radius threads can be formed via cutting or rolling threads. Most preferably, UNR threads will be rolled to combine the benefits of having a lower stress concentration and the compressive stresses previously mentioned. Most fastener specifications that were written for fatigue applications suggest UNJ threads for inch series parts and MJ threads for metric parts. UNJ and MJ threads have a larger thread root radius than normal inch and metric series fasteners. This improves fatigue life by reducing the stress concentration in the threads. UNJ and MJ threads are required to be called out on a print due to how uncommon they are in the fastener industry (excluding aerospace fasteners). Note also UNJ and MJ external threads require to mate with UNJ and MJ internal threads, respectively.
    • For more information on how fastener threads are produced, see Screw Thread Design.
  • Achieving and Maintaining Clamp Load
    Torque control is the most common and widely used method for installing parts around the world, so a degree of understanding of how to appropriately install parts using torque-tension relationships is necessary to increase the effectiveness of the installation torque.
    Due to lack of torque control, impact wrenches should be avoided when installing fasteners. Tools which allow torque control to ±2% are suggested. Ideally, all installations should have a final torque check with a regularly calibrated torque wrench.
    Maintaining clamp load is equally as important as obtaining the proper clamp load. There exist excellent chemical and mechanical locking mechanisms for vibration applications like (1) pre-applied and point-of-use threadlocker, (2) two piece wedge lock washers, (3) serrated disc spring washers, (4) serrated flange nuts/bolts, and (5) specialty locking threads. Disc spring washers (6) are excellent for thermal fluctuation applications.
    Please note that nylon insert locknuts, all-metal locknuts, internal/external tooth and split lock washers are not listed here as effective ways of preventing loosening or fatigue. For heavy vibration and/or fatigue applications, one of the six listed items in the previous paragraph is recommended.
  • Proper Mating Thread Material
    Mating internal threads must be capable of supporting the tensile and shear force required to fully seat the fastener to its required clamp load. If the material is too weak, stripping, relaxation, embedment, and an inability to achieve and maintain clamp load can occur, leading to fatigue.
    This does not mean the hardness of the internal threads needs to be equal to the hardness of the bolt. Internal thread hardness, typically, is lower than the bolt threads, yet it is still able to provide the proof load necessary to equal the tensile strength of the bolt. Take for example a 1/2 inch diameter SAE J429 Grade 8 HCS mated with an SAE J995 Grade 8 nut. The bolt has a core hardness requirement of HRC 33 to 39 while the mating nut has a core hardness of HRC 24 to 32. Yet the nut is able to provide a proof strength of 150,000 PSI which is equal to the minimum tensile strength of the bolt.
    A softer internal thread mating material also helps distribute the clamp load to the mating threads. The first engaged thread experiences an enormous amount of tension. The internal thread is designed to deform and potentially yield, which transfers some of the load to the adjacent internal thread, which in turn deforms and transfers to the next thread, and so on. The first few threads still support the majority of the load, but it is far less than if the internal threads were harder. This load distribution is critical for longer fatigue life and is also the reason why reusing nuts is not advised (because they have yielded).
  • Corrosion Resistant Material and Corrosion Control
    There are fatigue mechanisms which require corrosion to propagate a crack or where corrosion simply exacerbates the problem. In almost every application corrosion is unwanted and should be controlled. To do this, we can either select a material to make the fastener from that is more resistant to the environment the parts are going into, or we can use a coating system that will provide galvanic or barrier protection, or better yet, a combination of base material selection and coating system.
  • Avoid High-Strength Fasteners When Not Required
    Many people make the mistake of increasing fastener strength when a failure has occurred thinking that a stronger fastener will prevent a reoccurrence. A higher strength fastener offers the advantage of being able to produce higher clamp loads; however, without utilizing the higher clamp loads by increasing torque and using stronger internal threads, all we get is a harder more brittle material under the same cyclical loading. Endurance limits of a harder material may be higher, but this is not a solution to prevent the fatigue loading.
  • Routine Maintenance and Replacement
    If you know you have experienced fatigue failures in your application and you have tried everything above to solve the problem, you may need to put your application on a maintenance program. Perform periodic inspections with a torque wrench to check for joint loosening or relaxation. If this doesn’t work, start recording when each part is installed and when they fail. Once you know how long the parts will last, pick a percentage of this lifetime and replace all similar parts in regular intervals.

 
  • Expand All
  • Collapse All