1. Principles of Fatigue
1.1 Basics of fatigue failure
Fatigue failure is the progressive loss of integrity of a solid subjected to repeated or fluctuating loads. Even when the peak load is below the static strength of the material, cyclic stress and strain can cause microstructural damage. The process is typically gradual at first, but failure often occurs suddenly when a growing crack reaches a critical size for the applied loading.
A fatigue assessment aims to prevent premature cracking or fracture during the intended service life. In practice, this requires linking cyclic loading conditions to material response and to the stress state at the most critical locations.
1.2 S-N and strain-life concepts
Fatigue is commonly characterized using two complementary frameworks. The S–N approach relates stress amplitude (S) to the number of cycles to failure (N), using empirical curves obtained from tests. The strain-life approach instead relates cyclic strain amplitude to life, and it is especially relevant when plastic deformation occurs in each cycle.
The choice of approach depends on whether the component is expected to operate in the high-cycle, predominantly elastic regime or in the low-cycle regime with significant plasticity.
1.3 Crack initiation vs. crack growth
Fatigue life can be subdivided into two phases: crack initiation and crack growth. Crack initiation involves the development of microcracks at or near the surface or internal defects. Crack growth follows once a crack of sufficient size forms, allowing a stable increase in crack length under cyclic loading until final fracture.
Which phase dominates depends on materials, surface condition, geometry, loading spectrum, and inspection strategy. Some design approaches treat initiation and growth explicitly, while others incorporate both into an effective life model.
1.4 Typical fatigue regimes and failure modes
Common fatigue regimes include high-cycle fatigue (HCF), low-cycle fatigue (LCF), and sometimes very-high-cycle fatigue (VHCF) for extremely long lives. Failure modes vary with geometry and loading: surface-breaking cracks are frequent under bending or rotating loads; subsurface failures can occur in hardened or defect-sensitive materials; and multi-axial loading can produce mixed-mode cracking.
Failure mode mapping helps identify which model assumptions are appropriate and where to focus design changes.
2. Loading Characterization
2.1 Stress and strain spectra
Real service loads rarely repeat at a single constant amplitude. Loading characterization converts measured or simulated histories into a spectrum describing how often particular stress (or strain) ranges occur. The spectrum is then used in fatigue calculations to estimate damage accumulation across the entire loading history.
Accurate spectra are essential because fatigue damage often scales nonlinearly with load range, making the higher-stress events disproportionately influential.
2.2 Variable-amplitude loading
Variable-amplitude loading describes the general case where cyclic loads change over time due to operational variability, maneuvers, start-stop cycles, or environmental effects. The assessment must account for how different cycles interact in causing damage.
To manage complexity, engineers often transform histories into equivalent representations such as rainflow cycle counts or simplified block spectra, while ensuring conservatism consistent with uncertainty.
2.3 Mean stress and load ratio (R)
Cyclic stress states include both alternating and mean components. Mean stress can shift material response and affect crack driving force. The load ratio, commonly denoted R, is used to describe the relationship between minimum and maximum cyclic stresses.
Mean-stress corrections are incorporated through fatigue models that adjust life predictions to reflect tensile mean stresses as more damaging than compressive ones, subject to the specific formulation and calibration used.
2.4 Simplified load histories and equivalent cycles
When full spectra are unavailable or overly complex, simplified models represent the duty cycle using equivalent constant-amplitude cycles or a reduced set of cycle blocks. Equivalent cycle methods translate the variable-amplitude problem into a form that can be used with S–N or strain-life relationships.
The reliability of simplified histories depends on how they preserve the damaging portion of the spectrum and how uncertainty is handled in later safety-factor or probability-based steps.
3. Material Properties for Fatigue Design
3.1 Cyclic stress-strain response
Under cyclic loading, materials do not respond exactly as they do in monotonic tests. Engineers often need cyclic stress-strain properties, including how the material hardens or softens with repeated cycles. These properties enable strain-life calculations and improve the realism of low-cycle fatigue predictions.
For components operating near elastic conditions, cyclic properties may be approximated, but verification against test data is still important when precision affects acceptance.
3.2 Fatigue strength and fatigue ductility coefficients
For the strain-life framework, fatigue strength and fatigue ductility coefficients summarize how a material’s cyclic stress and strain relate to life. These coefficients are typically derived from experimental data across a range of strain amplitudes.
Using these coefficients without matching the material condition (heat treatment, surface condition, manufacturing route) can introduce significant error, especially for components sensitive to microstructural variations.
3.3 Endurance limits and approaches for non-ferrous materials
Ferrous materials are often modeled with an endurance limit concept, where life approaches very large numbers beyond a particular stress level. For non-ferrous alloys, the presence of a true endurance limit is less consistent, and engineers may use high-cycle extrapolations instead.
This difference influences long-life design decisions and the way acceptance criteria are set for very high cycle counts.
3.4 Effects of temperature and environment
Temperature can alter fatigue strength, stiffness, and deformation mechanisms, shifting the S–N or strain-life behavior. Environment also affects fatigue through corrosion and oxidation, which can reduce allowable stress ranges by promoting surface damage.
Design therefore requires material data under relevant temperature ranges and environmental exposure, or justified adjustments when such data are limited.
4. Stress Analysis for Fatigue
4.1 Determining nominal vs. local stresses
Fatigue assessments begin with an estimate of the nominal stress based on applied loads and basic geometry. However, cracks initiate at specific locations where the local stress state differs from the nominal value due to shape changes and constraints.
The engineering task is to determine the stress measure that correlates with fatigue damage—often using a combination of nominal analysis, geometric correction factors, and, when warranted, local stress extraction.
4.2 Stress concentrations and geometric discontinuities
Geometric discontinuities such as notches, holes, threads, fillets, and abrupt thickness changes amplify stresses. Under cyclic loading, these hotspots govern where damage begins.
Fatigue sensitivity to stress concentrations is generally higher than for static strength considerations, making careful geometry assessment a key step.
4.3 Notch effects and fatigue sensitivity
Notch effects include how stress distribution around a discontinuity influences local strain and crack initiation. The severity depends on notch radius, material properties, and loading mode. Some approaches use notch sensitivity factors to scale from theoretical stress concentration to an effective fatigue-relevant stress.
Selecting appropriate notch factors ensures that the fatigue model does not under- or over-estimate life due to overly optimistic or overly conservative local stress assumptions.
4.4 Finite element analysis for fatigue
Finite element analysis (FEA) is used to compute stresses in complex geometries that cannot be evaluated accurately with hand calculations. For fatigue, analysts often focus on elastic stress results (for high-cycle regimes) or on cyclic plasticity-capable methods (for low-cycle regimes).
When using FEA for fatigue, model credibility depends on mesh quality, boundary conditions, element formulations, and accurate extraction of stress parameters at critical locations.
5. Design Detail Considerations
5.1 Surface finish and residual stress
Surface condition strongly influences fatigue because cracks frequently initiate at the surface. Roughness can raise local stress intensification and act as a site for early crack nucleation. Residual stresses, such as those introduced by manufacturing or heat treatment, can either retard or accelerate cracking depending on their sign and magnitude.
Surface treatments and finishing processes are therefore integral to fatigue design rather than cosmetic considerations.
5.2 Corrosion and fretting effects
Corrosion fatigue arises when cyclic loading and chemical attack combine to reduce life relative to either effect alone. Fretting fatigue occurs at interfaces where small relative motions lead to local damage and crack initiation, often under bolted joints, press fits, or sliding contacts.
Mitigation includes protective coatings, lubrication, improved clamping, and design changes that reduce micro-slip.
5.3 Material defects and inclusions
Internal defects such as voids, pores, and inclusions can become fatigue crack initiation sites, particularly when surface quality is high. The influence depends on defect size distribution and stress level at the defect location.
Quality control during manufacturing—such as inspection and process controls—supports the fatigue design by narrowing variability and avoiding rare but damaging defect populations.
5.4 Welding and additive manufacturing considerations
Welded joints introduce geometrical features, residual stresses, and microstructural heterogeneity that can dominate fatigue performance. Fatigue-sensitive zones often include the weld toe, root region, or heat-affected zones. Appropriate weld detailing, post-weld treatments, and use of joint-specific fatigue categories are common mitigation steps.
Additive manufacturing can also affect fatigue through anisotropy, porosity, and surface irregularities. Fatigue assessments typically require material data representative of the build orientation and post-processing state.
6. Fatigue Damage Modeling
6.1 Goodman, Gerber, and Soderberg mean-stress corrections
Mean-stress corrections adjust fatigue predictions when the cyclic waveform includes a tensile mean component. Goodman, Gerber, and Soderberg are widely used forms, each offering different conservatism and sensitivity to mean stress.
Selection among these corrections is guided by the material system, available calibration data, and the desired balance between conservatism and realism.
6.2 Multiaxial fatigue criteria
Many components experience combined stress components rather than simple uniaxial loading. Multiaxial fatigue criteria translate a multi-component stress state into an equivalent parameter that can be used with uniaxial fatigue relationships or directly within damage models.
The choice of criterion depends on the dominant crack mechanisms and material behavior, and it should be supported by validation when critical.
6.3 Miner’s rule and damage accumulation
Miner’s rule provides a linear damage accumulation concept: damage is summed over cycle blocks based on the fraction of life consumed at each stress range. The total damage reaching unity is treated as indicative of failure.
Despite its simplicity, Miner’s rule is widely used because it integrates naturally with load spectra. Its accuracy depends on the applicability of linear accumulation to the material and spectrum; uncertainty may require safety factors or probabilistic treatment.
6.4 High-cycle vs. low-cycle fatigue transitions
The boundary between high-cycle fatigue and low-cycle fatigue is not fixed, but it is often identified by the onset of significant cyclic plasticity. Transition modeling matters because HCF and LCF predictions use different parameters (stress amplitude vs. strain amplitude) and different material behavior assumptions.
When components span the transition region, combined or carefully selected models are needed to avoid discontinuities in predicted life.
7. Crack Growth and Fracture Mechanics
7.1 Paris/E prandtl-type crack growth relationships
Fracture-mechanics-based fatigue models use relationships between crack growth rate and the stress intensity factor range experienced by a crack. The Paris law and related formulations express crack growth as a power function of the effective stress intensity range.
Such models enable life prediction when an initial crack size is assumed or estimated and when stable crack growth governs a significant portion of total life.
7.2 Threshold and instability considerations
Crack growth models often include a threshold below which growth is very slow, governed by material resistance to crack propagation. At the other extreme, instability occurs when the driving force exceeds a critical level, causing rapid fracture.
Accounting for both threshold behavior and instability conditions improves the fidelity of crack-growth-based life estimates, particularly for slow-growth designs and for final fracture prediction.
7.3 Characteristic crack size selection
To use crack growth models, engineers must decide on an initial crack size and whether to treat crack initiation separately. Choices may reflect detectability limits, manufacturing defect assumptions, or conservative “worst-case” sizes.
The characteristic crack size selection substantially affects results, which is why documentation of assumptions and alignment with inspection capabilities is emphasized.
7.4 Inspection intervals and damage tolerance concepts
Damage tolerance approaches plan for the possibility of crack growth and rely on inspections to detect cracks before they reach critical size. Inspection interval selection uses crack growth predictions, uncertainty, and inspection reliability.
This framework shifts design from purely preventing cracks to managing the risk that cracks might exist, making it especially relevant for safety-critical structures.
8. Life Prediction and Safety Factors
8.1 Deterministic life estimation
Deterministic life estimation produces a single predicted life value using specified material properties, stress calculations, and damage models. The result supports engineering decisions like acceptance testing, design iteration, and qualification.
Because inputs contain uncertainties, deterministic predictions typically incorporate safety factors or conservative assumptions to control the risk of underestimation.
8.2 Reliability-based and probabilistic methods
Reliability-based methods treat key parameters—loads, material properties, defect sizes, and model scatter—as random variables. Life is then described using probability of failure or confidence bounds rather than one fixed number.
Probabilistic frameworks can better capture variability and support optimized safety targets, but they require more input data and statistical modeling effort.
8.3 Uncertainty sources in fatigue predictions
Uncertainty arises from measurement errors in loading spectra, modeling approximations in stress analysis, variability in material fatigue properties, manufacturing variation, and limitations in chosen fatigue criteria. Environmental effects and degradation over time also contribute to uncertainty.
Quantifying dominant uncertainty sources helps target additional testing or tighter controls where they yield the greatest reduction in prediction risk.
8.4 Safety factors and acceptance criteria
Safety factors convert predictions into design limits by accounting for uncertainties and desired margins. Acceptance criteria specify allowable damage or required life under the defined loading and inspection context.
A sound acceptance framework is consistent across design, testing, and review, ensuring that qualification conclusions remain traceable to the underlying assumptions.
9. Fatigue Testing and Calibration
9.1 Specimen types and test setups
Fatigue testing uses standardized specimen geometries and fixtures that replicate relevant stress states as closely as practical. Test setups control load application methods, measurement systems, and environmental conditions.
When specimens cannot fully replicate component geometry, calibration or correlation is used to bridge the gap between test conditions and actual service details.
9.2 Obtaining S-N curves from experiments
S–N curves are obtained by conducting fatigue tests at multiple stress amplitudes and recording cycles to failure. Statistical treatment may be used to account for scatter and censored data when run-outs occur.
The resulting curves are then used by fatigue design models, with careful selection of curve characterization methods consistent with the intended design use.
9.3 Correlating test data with design details
Design details such as notches, weld profiles, surface treatments, and thickness effects can shift fatigue behavior relative to smooth specimens. Correlation uses notch factors, joint categories, or detail-specific calibration to align test-derived curves with component-relevant conditions.
Correlating data reduces reliance on purely generic assumptions and improves prediction credibility.
9.4 Verification and validation strategies
Verification confirms that analysis methods are implemented correctly and that modeled stresses align with expected results. Validation checks that the overall modeling approach predicts outcomes consistent with experiments for relevant geometries and loading cases.
Robust fatigue programs typically combine both, using a mix of qualification tests and targeted confirmatory trials.
10. Mitigation and Design Improvements
10.1 Reducing stress concentrations
Stress concentration reduction is a primary design lever. Changing fillet radii, smoothing transitions, repositioning holes, and improving load paths can lower local stress gradients and delay crack initiation.
Mitigation must balance fatigue improvements with other constraints such as weight, manufacturability, and stiffness requirements.
10.2 Improving surface treatment and finishing
Surface treatments aim to reduce roughness and to improve near-surface fatigue resistance. Options include polishing, controlled machining practices, coatings, and processes that produce beneficial compressive residual stresses.
Because surface state can evolve through service (e.g., via corrosion), mitigation often includes both immediate and durability considerations.
10.3 Shot peening and residual stress management
Shot peening introduces compressive residual stress near the surface, which can counteract tensile cyclic stress effects. Residual stress management also considers how subsequent machining, heat, or assembly processes may relax the compressive state.
Residual stress measurements and process qualification help ensure that peening results translate into improved fatigue performance.
10.4 Material selection and dimensional optimization
Material selection considers fatigue strength, ductility, corrosion resistance, and microstructural stability. Dimensional optimization adjusts thicknesses, stiffness, and geometric proportions to distribute stresses more favorably.
Even modest changes in geometry or material state can produce significant life gains when fatigue is governed by a small number of critical stress locations.
11. Practical Design Workflow
11.1 Step-by-step fatigue design procedure
A typical workflow starts by defining service loads and duty cycles, followed by identifying critical components and likely crack initiation sites. Next, engineers compute nominal stresses and apply geometric and detail corrections or local stress extraction. A fatigue model is selected based on expected fatigue regime.
Finally, life is calculated, safety factors or reliability targets are applied, and design changes are iterated until acceptance criteria are met.
11.2 Selecting analysis level and models
Analysis level ranges from simplified hand calculations to detailed local stress and crack growth models. Model selection depends on criticality, complexity of geometry, availability of loading data, and expected dominance of initiation or crack growth.
Selecting models that match the available evidence improves efficiency while keeping predicted risk within acceptable bounds.
11.3 Documenting assumptions and calculations
Fatigue assessments rely on assumptions about load representation, stress extraction methods, material data sources, and model suitability. Documentation typically includes the selected fatigue framework, mean-stress treatment, crack growth assumptions if used, and the basis for safety margins.
Clear traceability supports review, future maintenance, and updates when loads or operating conditions change.
11.4 Common pitfalls and troubleshooting checklist
Frequent pitfalls include using nominal stresses without appropriate detail correction, applying mean-stress corrections outside their calibration range, ignoring surface or manufacturing effects, and misrepresenting the loading spectrum. Another common issue is inconsistent unit handling or mismatched definitions of stress amplitude and life models.
A troubleshooting checklist typically verifies inputs, model consistency, stress extraction at critical locations, and alignment between test data and component details.
12. Standards, Guidelines, and Reporting
12.1 Overview of engineering fatigue standards
Engineering fatigue standards provide guidance for data selection, test methods, analysis procedures, and acceptance practices. They also define how to structure fatigue categories for details such as welded or machined joints.
Standards vary by industry and region, but their common purpose is to improve reproducibility and reduce ambiguity in design and qualification.
12.2 Required inputs and typical calculation outputs
Required inputs commonly include loading spectra or load cases, material fatigue properties (and cyclic stress-strain data when needed), geometric descriptors, and stress analysis results at critical locations. Typical outputs include predicted life (cycles or time), damage accumulation summaries, crack growth estimates where applicable, and safety factor comparisons.
For reliability-based designs, outputs may also include probability of failure or confidence bounds rather than single-life predictions.
12.3 Reporting conventions for fatigue assessments
Fatigue reporting typically includes the defined scope, assumptions, loading characterization method, model selection rationale, stress evaluation details, and results for each critical location. Reporting also documents uncertainties and the basis for any safety margins.
Consistent conventions facilitate cross-checking between design teams, reviewers, and test engineers.
12.4 Compliance and review considerations
Compliance and review focus on whether the assessment conforms to applicable standards and internal requirements. Review often includes validation of loading assumptions, verification that stress analysis is adequate, and scrutiny of model applicability and calibration sources.
Where gaps exist, review may request additional testing, tighter controls, or revised safety margins to ensure the final design meets its intended risk and performance goals.