What Is Metal Fatigue?
Metal fatigue is progressive damage caused by repeated or fluctuating stress or strain. A small crack can form, grow over many load cycles and eventually cause a part to break. Fatigue can occur below the metal’s static yield strength because deformation and damage concentrate locally, even while the whole part appears unchanged.
Why can repeated stress break metal below yield?
The stress calculated for a whole section can hide much higher local demand. A thread root, shoulder, weld toe, scratch or corrosion pit concentrates stress. Repeated microscopic slip can create irreversible damage there without making the entire component visibly yield. Even a smooth specimen can develop fatigue damage at microstructural features.
Local cyclic deformation creates a small crack at a vulnerable surface, microstructural feature or defect. An existing flaw can greatly shorten this initiation period.
Continued damaging cycles can extend the crack. Growth rate changes with crack size, loading, material and environment; a load cycle does not imply a fixed amount of advance.
The remaining section becomes unable to carry the load, or the crack reaches an unstable condition. Final separation can then happen rapidly.
A shaft can experience alternating bending as it rotates under a steady external force. Brackets can cycle with vibration, pipes with pressure changes, and constrained parts with heating and cooling. Fatigue therefore depends on the load history at the critical location, not simply on how old the metal is.
A fatigue failure may finish with an overload fracture. That final event does not erase the earlier crack-growth history. A single excessive load, sustained high-temperature creep and stress-corrosion cracking have different primary drivers, although more than one mechanism can act together.
Stress range, amplitude and mean stress
A maximum load alone does not describe a fatigue cycle. Record both extremes, their sequence and the number of repetitions. For one regular stress cycle:
Range and amplitude differ by a factor of two. Check which one a fatigue curve uses before entering a stress value. The stress-ratio expression requires a nonzero maximum stress.
Worked example: a cycle from 20 to 120 MPa has a range of 100 MPa, an amplitude of 50 MPa, a mean of 70 MPa and R ≈ 0.167. A fully reversed cycle from −120 to +120 MPa has the same maximum, but its amplitude is 120 MPa and R = −1.
These are different fatigue inputs. Their life cannot be compared from the maximum stress alone. Tensile mean stress can reduce resistance in many unwelded materials; welded-detail rules often account for residual stress differently.
How to read an S–N curve
An S–N curve relates a defined cyclic stress measure to cycles to a stated failure criterion. Under otherwise comparable conditions, lower cyclic stress generally corresponds to a longer fatigue life.
Read the axes first. S may mean stress amplitude or stress range; N is usually plotted on a logarithmic scale. Then check the material, heat treatment, surface, geometry, stress ratio, temperature, environment and statistical basis.
A runout means the test stopped before the failure criterion was reached. It is evidence that the specimen survived that test duration, not proof of unlimited life.
There is no universal number of cycles that causes fatigue. Some materials show an endurance limit under specified test conditions; other material–environment combinations do not. A polished-specimen curve should not be transferred directly to a corroded, notched or welded part.
High-cycle and low-cycle fatigue
High-cycle assessment commonly treats the overall cyclic response as mostly elastic and uses stress-life data. Low-cycle assessment focuses on appreciable cyclic plastic strain, often using strain-life data for severe start-stop or thermal cycling.
The transition depends on the local response and material. A fixed cycle count, such as 105, is not a universal dividing line.
Match the test to the question
ASTM E466 addresses force-controlled, constant-amplitude axial tests. ASTM E606/E606M addresses strain-controlled fatigue. ASTM E647 addresses fatigue crack-growth rates.
These specimen methods answer different questions. Representative component testing is still needed where geometry, joints or the service history cannot be transferred adequately from coupon data.
What do Paris’ law and Miner’s rule calculate?
Crack growth from a known or assumed flaw
A Paris-type relation estimates crack advance per cycle in the intermediate part of an appropriate crack-growth curve. Here a is crack length, N is cycle count, ΔK is stress-intensity-factor range, and C and m are fitted for specified test conditions. It is not valid across the entire growth history: small cracks, near-threshold behavior, large-scale yielding, environment and final instability need appropriate treatment.
Accumulating variable-amplitude cycles
Miner’s rule adds cycle fractions, where ni is the applied count at one level and Ni is the corresponding life from an applicable curve. D = 1 is a common idealized failure convention, not an exact physical clock. Sequence and interaction effects can matter. A damage sum is not a percentage of confirmed remaining life.
What changes a component’s fatigue life?
Material strength is only one input. The load path, local geometry and production condition determine whether a laboratory result describes the real part.
Loading and assembly
Load range, mean level, overloads and sequence all matter. Misalignment, poor balance, resonance, preload loss and fretting can create cyclic loads that were missing from the original calculation.
Geometry and surfaces
Sharp shoulders, keyways, holes and abrupt stiffness changes raise local demand. Tool marks, scratches, grinding damage and corrosion pits can become crack origins. Their location and direction matter as well as their size.
Material and process
Grade, heat treatment, microstructure, inclusions, porosity and surface condition affect initiation and growth. Residual tensile or compressive stresses from welding, machining or treatment also change the local response.
Environment and temperature
Corrosion can create pits and accelerate cracking. Temperature, dwell time and exposure may introduce creep or other damage alongside fatigue. A dry, room-temperature test does not establish performance in every service environment.
A real investigation: the WWVB tower pin
NIST traced a tower collapse to fatigue cracking in a replacement AISI 4140 steel insulator pin. The crack started at the fillet between its head and body. The investigation identified sharp radii, circumferential machining marks and pitting; drawings also lacked adequate finish and tolerance requirements.
The report explains why the replacement’s higher static strength did not offset poor detail and surface condition. Its recommendations addressed radii, tolerances, material and heat treatment together. Source: NISTIR 6619, conclusions and recommendations, printed pages 23–24.
Why welded joints need their own assessment
A weld’s toe, root, attachment geometry, alignment and residual stress can dominate its fatigue behavior. Many design methods use weld-detail categories and corresponding S–N curves instead of assigning life from the base metal’s tensile strength.
Follow the governing method’s stress definition and treatment of residual stress. Raising the alloy’s strength or increasing weld size does not automatically improve the fatigue detail. TWI explains the basis of these welded-detail rules.
For laser-welded parts, joint geometry and process condition remain part of the assessment. Our laser-welding HAZ guide explains the surrounding material changes; a narrow heat-affected zone alone is not evidence of fatigue life.
How can fatigue failures be reduced?
Reduce the cyclic demand, improve the vulnerable detail and verify the production condition. Inspection supports these controls but does not replace them.
- Measure the duty cycle. Include normal operation, startup, shutdown, vibration, impacts, thermal events and unusual loads. Define the required life and what counts as failure.
- Improve the load path. Review alignment, supports, balance, preload and resonance. Use suitable transitions and radii to reduce local stress concentration.
- Control the manufactured surface and joint. Specify critical geometry, finish, heat treatment, defect limits, weld condition and corrosion protection. Qualify any peening, dressing or other improvement process for the actual detail.
- Validate and control changes. Combine an applicable fatigue method with representative testing and inspection. Reassess changes to the material source, machining, heat treatment, welding, coating, assembly or service load.
Higher stiffness may reduce deflection but can also change where load is carried. Higher static strength does not by itself establish fatigue resistance. Keep these requirements separate when reviewing stiffness and strength.
How are fatigue cracks detected?
A visual check can miss a small, tight crack. Choose nondestructive testing (NDT) for the material, expected crack orientation and location, surface condition and available access.
No reportable indication means none was detected within the procedure’s capability and coverage. It does not prove that every crack is absent. For a fatigue-critical part, inspection intervals need a basis in detectable crack size, growth behavior and the critical condition.
On a narrow screen, scroll the table sideways.
| Method | Useful for | Important limit |
|---|---|---|
| Visual examination | Accessible damage, corrosion, fretting and visible cracks | Tight, hidden or subsurface cracks may not be visible. |
| Liquid penetrant | Surface-breaking cracks in suitable nonporous materials | Needs a clean, accessible surface; cannot reveal a fully buried crack. |
| Magnetic particle | Surface and near-surface cracks in ferromagnetic materials | Not suitable for aluminum or copper; magnetization direction affects detection. |
| Eddy current | Surface and near-surface cracks in electrically conductive materials | Depth sensitivity depends on material and frequency; probe, geometry, coatings and crack orientation affect response. |
| Ultrasonic testing | Internal or surface-connected flaws with suitable sound paths | Access, calibration and flaw orientation govern detection and sizing capability. |
FHWA’s inspection resources describe penetrant, magnetic-particle and eddy-current testing, including practical limitations. A qualified procedure must represent the part being inspected.
If a crack is suspected in a safety-critical or load-bearing part, take it out of service safely and obtain a qualified assessment before further loading or repair. Surface grinding, stop-drilling or adding a weld does not by itself establish that the crack and its cause have been addressed.
What a fracture surface can reveal
A fatigue fracture may contain an origin, a progressive-growth region and a rougher final-fracture region. Curved progression bands, often called beach marks, or multiple-origin steps can help an analyst reconstruct the sequence. These features are not present in every fatigue failure.
Beach marks can reflect changes in loading or environment. They should not be counted as individual service cycles. Microscopic striations, where present, also need interpretation in the context of the material and loading.
Confirm the mechanism using the fracture, material, geometry and operating history together. For welded parts, compare the possible mechanisms in our weld crack types guide.
Preserve the evidence before cleaning
Photograph the assembly, orientation and fracture faces before moving parts where safe to do so. Retain matching fragments, material and process records, operating logs, load or vibration data, and recent maintenance history.
Protect fracture faces from rubbing and contact. Do not wire-brush, blast, polish or force the halves together before a failure analyst decides how to examine them. Cleaning for an inspection and preserving a fracture for analysis are different tasks.
Calling the mechanism “fatigue” is only the beginning. The investigation must explain why the damaging cycles and vulnerable location existed, then verify that the corrective action addresses both.
Sources and test references
- NASA — Fatigue Testing: cyclic plasticity, initiation, stress-life and strain-life testing.
- NIST — Failure Analysis of the WWVB Tower: the documented pin failure and recommendations.
- TWI — Fatigue failure and prevention: crack development and the role of weld details.
- TWI — Fatigue design rules for welded structures: detail categories, residual stress and cumulative-damage limitations.
- ASTM: E466-21, E606/E606M-21 and E647-24: public scope pages for force-controlled, strain-controlled and crack-growth tests.
- FHWA — Evaluation of NDE Technologies for Fatigue Crack Detection: fatigue-crack inspection and monitoring capabilities.
Planning a joint that will see repeated loading?
Share the drawing, material condition, service loading and inspection requirements with Oceanplayer Laser to discuss representative welding samples.