Material Fatigue and Fracture

Material fatigue and fracture describe how engineering components fail when stress, flaws, environment, and time interact. A part can survive a single static load and still fail after repeated cycles. A high-strength material can still fracture suddenly if it contains a crack of sufficient size. A smooth specimen can look safe in a handbook table while the real component fails from a notch, weld toe, corrosion pit, fretting scar, or machining mark.

Fatigue and fracture are therefore not separate from design. They are central to aircraft structures, rotating shafts, bridges, pressure equipment, ships, wind turbines, rail components, welded frames, biomedical implants, springs, fasteners, and machines exposed to vibration or variable load.

The core engineering question is:

Can the material and geometry tolerate the expected stress history, flaws, environment, and inspection interval without crack initiation, unstable crack growth, or unacceptable deformation?

The answer requires a design basis, not only a material property. Fatigue and fracture decisions should state the load spectrum, local detail, material condition, environment, inspection capability, and consequence of failure.

Fatigue Design Basis

A practical fatigue review should define the following inputs before selecting formulas:

Input	Why it matters	Typical evidence
Local stress range	Fatigue initiates at local cyclic stress, not only nominal load.	Stress analysis, strain gauge data, hot-spot stress, detail category.
Mean stress	Tensile mean stress can accelerate crack opening and growth.	Load history, residual stress assumption, Goodman or similar screen.
Cycle count and spectrum	Damage depends on repeated cycles and sequence.	Duty cycle, rainflow counting, service logs, measured strain history.
Surface and notch condition	Small details often control initiation life.	Roughness, weld profile, notch radius, corrosion pit, machining route.
Material and process state	Toughness and fatigue strength depend on product form and heat treatment.	Certificate, heat treatment, hardness, microstructure, NDE record.
Inspection capability	Damage tolerance depends on what flaws can be found before fracture.	Method, access, detectable flaw size, interval, probability of detection.

This basis is what separates a defendable fatigue assessment from a handbook lookup.

Static strength is not enough

Static strength checks compare stress with yield strength, ultimate tensile strength, or another strength limit. Those checks are necessary but not sufficient. Fatigue failure can occur at stresses below yield strength if the stress repeats many times. Fracture can occur below nominal strength if a crack concentrates stress at its tip.

The difference matters because many service loads are cyclic:

• rotating shafts experience alternating bending;
• aircraft wings experience gust and pressurization cycles;
• bridges experience traffic cycles;
• pressure vessels experience pressure cycles;
• ships experience wave-induced stress cycles;
• welded frames experience vibration;
• implants experience repeated body motion.

For these systems, the maximum stress from one event is only part of the story. Engineers also need stress range, mean stress, cycle count, surface condition, notch severity, residual stress, crack size, and environment.

Fatigue mechanism

Fatigue usually develops in three stages. First, a small crack initiates at a local stress raiser. Common initiation sites include scratches, inclusions, pores, corrosion pits, weld toes, sharp fillets, threads, keyways, and surface damage. Second, the crack grows incrementally under repeated loading. Third, the remaining section becomes too small or the crack becomes too large, and final fracture occurs.

This staged process explains why fatigue can be difficult to detect. Most of the component may still look intact while a crack grows at one critical location. Once the crack reaches a critical size, final fracture may be rapid.

Fatigue resistance depends on:

• stress amplitude and mean stress;
• stress concentration and notch sensitivity;
• surface finish and residual stress;
• material microstructure and heat treatment;
• size and geometry;
• temperature and corrosion environment;
• loading frequency and sequence;
• inspection method and acceptance criteria.

No single material property captures all of these effects.

Stress-life design

The stress-life approach uses an S-N curve to relate cyclic stress amplitude to number of cycles to failure. It is most useful for high-cycle fatigue, where cyclic plastic strain is small and the material response is mostly elastic.

An S-N curve is built from fatigue tests at different stress amplitudes. The vertical axis is stress amplitude or stress range, and the horizontal axis is cycles to failure. For some steels, the curve approaches an endurance limit below which fatigue failure is not expected under the test conditions. Many non-ferrous alloys do not have a true endurance limit, so fatigue strength is specified at a chosen number of cycles.

Real components are not polished laboratory specimens. The design S-N curve or endurance limit may need correction for surface finish, size, loading mode, temperature, reliability, stress concentration, mean stress, and environment. A polished rotating-beam specimen value should not be applied directly to a welded, corroded, notched, full-size component.

Mean stress and the Goodman criterion

Cyclic loading is described by maximum stress, minimum stress, stress amplitude, mean stress, and stress ratio:

Tensile mean stress usually reduces fatigue life because it keeps cracks more open during the cycle. Compressive mean stress can be beneficial, although designers should be cautious about relying on it when residual stress may relax.

The modified Goodman criterion is a common screening rule:

where S_e is endurance limit or finite-life fatigue strength, \sigma_{UTS} is ultimate tensile strength, and N is a design factor. Goodman analysis is simple and transparent, but it is not universal. It is mainly a preliminary design tool for metallic components under stress-life fatigue assumptions.

Variable amplitude loading

Real components rarely see one constant stress amplitude. Loads may include startup cycles, overloads, vibration, temperature changes, pressure pulses, braking events, wave loading, gust loading, or random spectra.

A common screening method for variable amplitude fatigue is the Miner rule:

where n_i is the number of cycles applied at stress level i, and N_i is the number of cycles to failure at that level from the S-N curve. Failure is often assumed when D approaches 1.

Miner damage is useful because it is simple, but it ignores sequence effects, overload retardation, crack closure, residual stress changes, corrosion, and nonlinear damage accumulation. For critical components, load spectra should be measured or justified, and the damage model should be validated against relevant data.

Worked Miner Damage Screen

Suppose a shaft detail is checked against an S-N curve and the expected annual spectrum is simplified into two stress blocks:

Stress block	Applied cycles n_i	Cycles to failure N_i
Moderate vibration	100,000	10,000,000
Startup transient	20,000	500,000

Miner damage for one year is:

If this simplified spectrum repeats unchanged, the linear damage screen suggests about:

This is not a life guarantee. It assumes the S-N data, local stress, surface condition, environment, and sequence effects are valid. It also ignores crack detection and repair. The value of the calculation is to expose which stress block dominates damage and where measurement or design improvement matters most.

Load spectra and service evidence

Fatigue analysis is only as strong as the load spectrum behind it. Design spectra may come from standards, simulations, strain-gauge measurements, accelerometers, pressure histories, sea-state data, traffic counts, flight cycles, startup records, or rotating-machine vibration data. Each source has uncertainty, and each may miss rare events that control damage.

Service evidence should be checked against the original assumptions. A bridge may carry heavier vehicles than expected. A machine may start more often after a process change. A ship may operate in rougher routes. A wind turbine may experience changed control settings. A biomedical implant may see patient activity outside the design envelope. These changes can turn a conservative fatigue case into an optimistic one.

For critical assets, monitoring and inspection results should feed back into the analysis. Measured strain ranges, crack findings, repair records, corrosion maps, and failure reports can justify longer intervals, trigger shorter intervals, or reveal that the model is missing an important detail.

Fracture mechanics

Fracture mechanics starts from a different assumption: a flaw or crack may already exist. The question becomes whether that flaw will grow and whether it can remain stable until detected or repaired.

For linear elastic fracture mechanics, the stress intensity factor is often written:

where Y is a geometry factor, \sigma is nominal stress, and a is crack size. Fracture occurs when the stress intensity reaches a critical value:

where K_{IC} is plane-strain fracture toughness. This relation links stress, crack size, geometry, and material toughness. It also shows why high strength alone is not enough. A strong but low-toughness material can fail from a relatively small flaw.

Fatigue crack growth

Once a crack exists, each load cycle can extend it. The driving quantity is usually the stress intensity range:

In the stable crack-growth region, the Paris-Erdogan law is commonly used:

where da/dN is crack extension per cycle, and C and m are material and environment dependent constants. Integrating this relation estimates the number of cycles for a crack to grow from an initial detectable size to a critical size.

Crack growth analysis is central to damage-tolerant design. It helps set inspection intervals, allowable flaw sizes, retirement lives, proof-test requirements, and non-destructive testing sensitivity. The method depends strongly on crack geometry, load spectrum, environment, residual stress, closure, threshold behaviour, and fracture toughness.

Ductile, brittle, and environment-assisted fracture

Fracture behaviour depends on material and service condition. Ductile fracture involves significant plastic deformation and often stable tearing before final failure. Brittle fracture can occur with little visible deformation, especially in high-strength materials, ceramics, low-temperature service, thick constrained sections, or embrittled metals.

Environment can shift behaviour. Corrosion fatigue, stress corrosion cracking, hydrogen embrittlement, high-temperature oxidation, irradiation, and liquid-metal embrittlement can reduce fatigue life or fracture resistance. A design that is safe in dry laboratory air may not be safe in saltwater, sour service, humid atmosphere, high temperature, or biological fluid.

Welds, notches, and surface condition

Fatigue and fracture are often controlled by local details rather than bulk material strength. Weld toes, weld roots, undercut, lack of fusion, sharp transitions, keyways, threads, holes, fretting interfaces, and surface scratches can dominate fatigue life.

Stress concentration is especially important because fatigue cracks usually initiate at local peaks. A small geometric change can shift a component from long life to short life. Improvements may include larger fillets, smoother machining, polishing, shot peening, compressive residual stress, better weld profiling, improved alignment, reduced roughness, protective coating, or lower stress range.

For welded structures, nominal stress, structural stress, hot-spot stress, and local notch stress methods may give different interpretations. Weld quality, residual stress, detail category, inspection class, and service environment must be considered together.

Inspection and damage tolerance

Inspection is part of fatigue and fracture design when cracks cannot be ruled out. Non-destructive testing methods include visual inspection, dye penetrant testing, magnetic particle testing, ultrasonic testing, radiography, eddy-current testing, acoustic emission, and structural health monitoring.

The inspection method must be matched to flaw type, material, geometry, access, surface condition, and required detection size. A damage-tolerance assessment should compare:

1. the largest flaw that might be missed;
2. the crack-growth rate under expected loading;
3. the critical crack size for fracture;
4. the inspection interval and probability of detection;
5. the consequence of failure.

This is more defensible than assuming cracks are absent simply because a part passed an initial inspection.

Inspection acceptance should be quantitative. The assessment should state the assumed initial flaw size, the largest flaw that may be missed, the critical crack size, and the crack-growth life between them. If the inspection method cannot reliably detect the required flaw size at the required location, the design is not damage tolerant even if the stress calculation looks acceptable.

Manufacturing, repair, and residual stress

Manufacturing route changes fatigue and fracture behaviour. Casting porosity, forging flow, additive-build defects, machining marks, grinding burn, heat-treatment variation, weld residual stress, coating defects, and surface roughness can all shift initiation life or crack-growth behaviour. A material certificate alone may not capture the local condition that controls failure.

Repair also needs fatigue review. Weld repair, blend-out grinding, drilled stop holes, sleeve installation, cold expansion, peening, replacement fasteners, and local heat treatment can reduce or improve life depending on geometry, residual stress, inspection quality, and load path. A repair that restores static strength may still leave a notch, hard zone, tensile residual stress, or hidden crack.

Engineering control should therefore link drawings, process specifications, inspection criteria, accepted defect size, and repair procedures. Fatigue-critical parts need manufacturing evidence that matches the assumptions used in the life calculation.

Validation evidence can include coupon testing, component fatigue testing, strain measurement, fracture-toughness data, crack-growth data, NDE records, failure analysis, and service inspection trends. Acceptance criteria should define the governing detail, stress range, assumed flaw size, inspection interval, environmental condition, and permitted repair. Otherwise the fatigue assessment cannot be maintained when the component, process, or duty changes.

Practical design workflow

A practical workflow is:

1. Define load cases, load spectrum, environment, temperature, service life, and consequence of failure.
2. Identify likely initiation sites: notches, welds, holes, contacts, threads, scratches, corrosion pits, and inclusions.
3. Estimate local stress amplitude and mean stress at critical locations.
4. Choose an S-N, strain-life, or fracture-mechanics approach based on stress level, crack assumptions, and required confidence.
5. Apply surface, size, loading, reliability, notch, and mean-stress corrections where appropriate.
6. Check static yielding and ultimate strength separately from fatigue.
7. Evaluate crack growth and fracture toughness when flaws may exist or inspection is part of the safety case.
8. Include environment, residual stress, manufacturing process, and material variability.
9. Validate with test data, service data, inspection results, or conservative design rules.
10. Document assumptions so later changes in material, surface finish, process, or load spectrum trigger review.

Common mistakes

Common mistakes include using static strength as a fatigue limit, applying polished-specimen S-N data directly to real parts, ignoring mean stress, treating the Miner rule as exact, using tensile strength instead of fracture toughness, and checking stress concentration only for static strength while ignoring crack initiation.

Another serious mistake is separating fatigue analysis from inspection. If cracks may exist, the design must consider detectable flaw size, crack-growth life, critical crack size, and inspection interval. A component is not damage tolerant unless the inspection and maintenance plan are part of the analysis.