Residual Stress in Welded Joints and Its Impact on Fatigue Life
Residual stress in welded joints is one of the most consequential and frequently underestimated factors governing the structural performance of fabricated components under cyclic loading. Welding is inherently a localised thermal process — intense heating followed by rapid, non-uniform cooling — and this thermal history introduces a self-equilibrating internal stress field that is superimposed on any subsequent service loading. When the structure operates under cyclic conditions, these residual stresses interact with applied loads to control fatigue crack initiation and propagation, often reducing fatigue life to a fraction of what the same component would achieve in an unwelded state.
A particularly important — and counterintuitive — consequence of welding residual stress is that it effectively eliminates the fatigue benefit of higher-strength steel. In unwelded specimens, fatigue endurance increases with yield strength. In welded joints, this relationship breaks down entirely: the dominant influences on fatigue performance are weld geometry, stress concentration at the weld toe, embedded defects, and the magnitude of tensile residual stress — not the base metal tensile strength. This is why structural fatigue design codes (IIW, BS 7608, ASME Section VIII) define S-N design curves by joint class rather than material grade, and why approximately 80% of mechanical failures in engineering structures are attributed to fatigue initiating at welded connections.
This article provides a comprehensive treatment of residual stress in welded joints: how it forms during the welding thermal cycle, how it is resolved into longitudinal and transverse components, how it interacts with fatigue crack mechanics, what factors govern its magnitude and distribution, and — critically — what engineering methods are available to measure, reduce, and design around it. Whether you are responsible for fatigue assessment of offshore structures, pressure vessels, pipelines, or any cyclically loaded welded fabrication, this guide provides the technical foundation you need.
The Welding Thermal Cycle and Origin of Residual Stress
To understand residual stress in welds, it is necessary to trace the sequence of events during and after welding. The arc or heat source generates a highly localised thermal field that moves through the joint. Material immediately adjacent to the arc is heated to temperatures well above the melting point, while material just a few millimetres away remains at near-ambient temperature. This extreme thermal gradient is the root cause of residual stress.
Heating Phase
As the arc passes, weld metal and the immediately adjacent heat-affected zone (HAZ) expand. At low temperatures, this expansion is accommodated elastically. Above a critical temperature — roughly 600°C for carbon steels, where the yield strength drops to near zero — the material cannot sustain elastic stress and undergoes plastic deformation. The hot, low-strength weld metal is plastically compressed by the cooler, stiffer surrounding material. This plastic compression is the starting point from which residual tension develops on cooling.
Cooling and Solidification Phase
As the weld pool solidifies and the heat source moves on, the weld metal begins to cool and tries to contract. But the surrounding base material — which experienced lower peak temperatures and has already partly recovered its stiffness — restrains this contraction. The result is that the weld metal is placed in tension, and the surrounding base material is placed in compression to maintain force equilibrium. In multi-pass welds, this cycle repeats with each pass, and the cumulative residual stress state reflects the entire thermal history of the joint. The final residual stress distribution is a complex function of the heat input, joint geometry, number of passes, interpass temperature, and mechanical restraint.
Metallurgical Phase Transformations
In carbon and alloy steels, solid-state phase transformations during cooling further modify the residual stress state. The austenite-to-martensite transformation at the Ms temperature involves a volumetric expansion of approximately 2 to 4%. This expansion can partially or fully relieve tensile residual stress in the weld metal — or even generate compressive stress if the transformation occurs at sufficiently low temperature. This principle is deliberately exploited in low transformation temperature (LTT) consumables developed specifically to improve the fatigue performance of welded joints. For ferritic steels with conventional filler compositions, the transformation effect is modest; for high-strength martensitic welds or duplex stainless steels, it can be significant.
Residual Stress Components — Longitudinal and Transverse
Residual stresses in welded joints are conventionally resolved into two principal components relative to the weld axis: longitudinal residual stress (parallel to the weld direction) and transverse residual stress (perpendicular to the weld direction). Understanding both components — their origin, magnitude, and distribution — is essential for fatigue assessment.
Longitudinal Residual Stress
Longitudinal residual stress acts parallel to the weld run. Its primary source is the restraint of longitudinal contraction of the weld metal and HAZ as they cool. Because the weld bead runs along its length in an already solidified and cooler restraining medium, the longitudinal contraction is highly constrained. As a result, tensile longitudinal residual stress at or near the weld centreline typically reaches yield strength magnitude in structural steels — values of 350 MPa or higher are common in mild steel welds, and proportionally higher in higher-strength steels. The distribution across the plate width shows a sharp tensile peak at the weld, dropping to compressive values at increasing distance into the base material, and returning to zero at the plate edges. In wide plates with multiple weld passes, the compressive zone can extend several times the weld width from the centreline.
Transverse Residual Stress
Transverse residual stress acts perpendicular to the weld axis. It arises from the transverse shrinkage of the weld bead and HAZ as they cool, restrained by the material ahead and behind the arc path. The magnitude of transverse residual stress is generally lower than the longitudinal component — typically 30 to 60% of yield strength — but its significance for fatigue is disproportionately high. This is because it acts normal to the weld toe and therefore directly perpendicular to the plane of fatigue cracks that initiate there. A stress component normal to the crack plane drives Mode I opening and is the dominant fatigue crack driving force. For this reason, transverse tensile residual stress at the weld toe is recognised as the most fatigue-critical component of the residual stress field, even though it is not the largest in absolute magnitude.
Fatigue Crack Initiation at the Weld Toe
Fatigue failure in welded structures overwhelmingly initiates at the weld toe — the junction line between the weld reinforcement face and the base metal surface. This location concentrates three adverse conditions simultaneously, and understanding their interaction is central to any fatigue assessment or improvement strategy.
Geometric Stress Concentration
The transition from the weld reinforcement to the base plate surface creates a notch effect. The severity of this notch — quantified by the stress concentration factor Kt — depends on the weld toe angle, toe radius, and reinforcement height. A sharp, undercut weld toe with a small toe radius (r < 0.5 mm) can produce Kt values of 3 to 5 for transverse loading, meaning the local stress at the toe is three to five times the nominal stress in the plate. Even a well-executed weld with a smooth toe transition will exhibit Kt values of 1.5 to 2.5. This geometric notch effect accelerates fatigue crack initiation by concentrating cyclic plastic strain at the toe.
Metallurgical Notch Effect
The weld toe also coincides with the fusion boundary between the base metal and the weld metal, passing through a HAZ that has undergone complex microstructural changes. In carbon and low-alloy steels, the coarse-grained HAZ immediately adjacent to the fusion line has elevated hardness, reduced toughness, and potentially elevated hydrogen content from the welding process — conditions that promote early fatigue crack nucleation. Weld slag inclusions, cold laps, or microscopically sharp geometric discontinuities at the toe act as pre-existing crack-like defects from which fatigue cracks can propagate without any initiation period. See the mechanical testing guide for an explanation of how Charpy impact testing and hardness surveys map these HAZ characteristics.
Peak Tensile Residual Stress
The weld toe is also the location of peak transverse tensile residual stress. This residual tension acts on the same plane as the potential fatigue crack — perpendicular to the applied cyclic load in a transversely loaded butt weld — and increases the mean stress at the crack tip. The consequence is an elevation of the effective stress ratio R (R = σmin / σmax). At high R, the crack remains open for a larger fraction of the load cycle, crack closure is reduced, and the effective stress intensity factor range ΔK driving crack growth is increased. The combined effect of geometric stress concentration, metallurgical notch, and tensile residual stress makes the weld toe a consistently preferred fatigue crack initiation site regardless of applied load direction or joint type.
Fracture Mechanics of Residual Stress and Fatigue Crack Growth
The mechanics of how residual stress accelerates fatigue crack growth can be described quantitatively using linear elastic fracture mechanics (LEFM). This framework underlies both design code fatigue assessments and fitness-for-service evaluations of cracked welded structures.
Paris Law and the Role of Mean Stress
Fatigue crack growth rate in metals under constant amplitude loading is described by the Paris-Erdogan equation:
Stress Intensity Factor for Weld Toe Cracks
For a semi-elliptical surface crack at the weld toe, the stress intensity factor is expressed as K = Y · σ · √(πa), where Y is a geometry correction factor accounting for the crack aspect ratio, free surface effects, and the weld toe stress concentration. The total K includes contributions from: the applied nominal stress, the weld toe geometric stress concentration (Mk factor), and the residual stress distribution across the crack plane. In fitness-for-service assessments following BS 7910 or API 579, these three contributions are superimposed to calculate the total crack driving force and compare against the material fracture toughness to determine whether a detected flaw is critical or tolerable.
Threshold and Short Crack Behaviour
Paris law describes steady-state crack growth. Below a threshold ΔK value (ΔKth), crack growth effectively arrests. In unwelded steel, this threshold is exploited — the endurance limit corresponds to applied stress levels that keep ΔK below ΔKth. However, tensile residual stress raises the effective R ratio such that ΔKth itself decreases (threshold is lower at high R). In the limit of yield-magnitude tensile residual stress (R approaching 1.0), ΔKth may approach zero — meaning that in as-welded joints under any cyclic loading, there is technically no fatigue threshold. This is why IIW and ASME fatigue design curves for as-welded joints do not include an endurance limit: they continue to slope downward at all cycle counts.
Factors Governing Residual Stress Magnitude and Distribution
The magnitude and spatial distribution of residual stress in a welded joint are not fixed by material alone. They depend on a combination of process, geometry, material, and restraint variables, which the fabrication engineer can influence through design and procedure choices.
| Factor | Effect on Residual Stress | Influence Level | Design/Procedure Lever |
|---|---|---|---|
| Heat input (kJ/mm) | Higher heat input widens HAZ and plastic zone; generally increases residual stress magnitude and extent | High | Optimise interpass temperature and travel speed |
| Degree of mechanical restraint | Higher restraint (fixed fixtures, thick plates, adjacent stiff members) increases tensile RS magnitude | High | Design joint access; back-step sequencing |
| Joint geometry and thickness | Greater thickness and multi-pass build-up increase RS complexity and magnitude | Medium | Minimise weld volume; select narrow groove |
| Number of passes and sequence | Each pass re-heats and modifies prior RS; final passes near the surface govern surface RS state | Medium | Optimise pass sequence; cap pass temperature control |
| Preheat and interpass temperature | Higher preheat reduces cooling rate and thermal gradient, modestly reducing RS | Medium | Follow P91 and CE-based preheat requirements |
| Solid-state phase transformation (Ms) | Low Ms temperature causes volume expansion on cooling, generating compressive RS in weld | Process-dependent | LTT consumables for fatigue-critical joints |
| Post-weld heat treatment (PWHT) | Reduces peak RS by 70–90% through plastic relaxation at elevated temperature | High — mitigating | Mandatory for pressure vessels; recommended for fatigue-critical structures |
| Welding process type | SAW produces higher heat input and deeper residual stress zones; GTAW produces narrower, lower RS zones | Medium | Select process appropriate to material and wall thickness |
Residual Stress Measurement Techniques
Accurate characterisation of residual stress requires appropriate measurement techniques. The choice of method depends on the required spatial resolution, depth of measurement, accessible geometry, and whether the component can be sectioned. All measurement techniques measure strain (or lattice strain), from which stress is calculated using the elastic constants of the material.
X-Ray Diffraction (XRD)
XRD measures the lattice plane spacing in crystalline materials using Bragg’s law. Residual stress shifts the peak diffraction angle relative to the unstressed reference; the shift magnitude, combined with elastic constants, gives surface residual stress. XRD is non-destructive and highly accurate but is limited to the near-surface region (penetration depth < 20 μm for conventional Cu Kα radiation in steel). Electropolishing layer-by-layer allows depth profiling at the cost of material removal. Laboratory and portable (Mx²) XRD systems are both used in industrial inspection.
Neutron Diffraction
Neutron diffraction uses the same principle as XRD but with neutrons rather than X-rays. Neutrons penetrate tens of millimetres into steel, enabling non-destructive through-thickness residual stress mapping with sub-millimetre spatial resolution. This technique is used extensively in research to validate finite element models of welding residual stress and to characterise stress in thick-walled pressure vessel welds. Access requires a neutron spallation or reactor source, so it is primarily a research tool rather than an industrial inspection method.
Hole-Drilling Method (ASTM E837)
Hole drilling is the most widely used semi-destructive method for industrial residual stress measurement. A small hole (typically 1.8 mm diameter, 2 mm depth) is drilled at the measurement point while strain gauges on a rosette bonded adjacent to the hole record the strain relief. The residual stress is back-calculated from the relieved strains using established influence coefficients. ASTM E837 defines the procedure for both uniform and non-uniform stress distributions. The method is applicable to any accessible flat surface and provides biaxial stress data. Its semi-destructive nature is acceptable for most structural components and pressure vessels.
Contour Method
The contour method provides a full 2D map of one stress component on a cross-section. The part is carefully cut by EDM wire along the plane of interest; the surfaces spring back as residual stress is released. The surface profile (contour) is measured by CMM or laser scanning, and the residual stress field is computed by applying the measured displacements as boundary conditions in a finite element model. This method is particularly powerful for characterising through-thickness stress distributions in thick-section welds and is increasingly used in the nuclear and offshore industries.
Mitigation Strategies for Residual Stress in Fatigue-Critical Welds
A range of engineering methods can reduce, redistribute, or counteract tensile residual stress in welded joints. These methods fall into three categories: thermal stress relief, mechanical surface treatment, and process modification. The most appropriate choice depends on joint geometry, accessibility, material type, service environment, and design code requirements.
Post-Weld Heat Treatment (PWHT)
PWHT is the most established and effective method for residual stress relief in welded fabrications. The component is heated uniformly to a temperature at which the material yield strength is reduced sufficiently to allow plastic flow and stress relaxation — typically 580 to 650°C for carbon and low-alloy steels, held for one hour per 25 mm of plate thickness, then slow cooled. Peak residual stress can be reduced by 70 to 90%, transforming the near-yield tension at the weld into relatively benign low-level stress. PWHT is mandatory for pressure vessels under ASME Section VIII and for pipework under certain ASME B31.3 conditions. For structural fabrication, PWHT significantly improves fatigue performance and shifts the applicable S-N design curve to a higher (less conservative) class in some code frameworks. The welding of P91 creep-resistant steels is a particularly critical application where PWHT requirements are precise and mandatory.
Peening Methods — Shot Peening, Hammer Peening, Ultrasonic Impact Treatment
Mechanical peening introduces compressive residual stress in the surface layer by plastic deformation, counteracting the tensile residual stress from welding in the fatigue-critical surface zone. Shot peening bombards the weld toe region with hard spherical media; hammer peening uses a pneumatic tool to plastically deform the toe; ultrasonic impact treatment (UIT) uses ultrasonically vibrating pins. All three methods achieve similar goals: they introduce compressive residual stress to a depth of 0.5 to 2 mm, smooth the weld toe geometry (reducing Kt), and close any micro-defects at the surface. Fatigue life improvements of 50 to 200% above the as-welded condition have been demonstrated in structural steel joints. IIW guidelines permit a fatigue class improvement of 2 to 3 categories for joints treated with effective peening.
Weld Toe Grinding and TIG Dressing
Weld toe grinding removes the geometric notch at the weld toe using a burr or disc grinder. By increasing the toe radius from typically < 0.5 mm to > 1 mm, the stress concentration factor Kt is substantially reduced. TIG dressing (remelting the weld toe surface with an autogenous GTAW arc) achieves a similar geometric improvement while simultaneously producing a small compressive residual stress from the rapid cooling of the remelted zone. Both methods are relatively straightforward to apply in production and can improve the applicable fatigue class by 1 to 2 categories. Grinding depth must be controlled to avoid undercutting into the base metal beyond 0.5 mm, which would introduce a new stress raiser. See the TIG/GTAW welding guide for parameters relevant to toe dressing.
Low Transformation Temperature (LTT) Consumables
LTT consumables are a process-level approach to residual stress management. By alloying the filler metal so that the martensitic transformation temperature (Ms) is reduced to approximately 200 to 250°C, the volume expansion accompanying the austenite-to-martensite transformation occurs at low temperature when the weld metal has significant stiffness. This transformation expansion counteracts the thermal contraction, generating compressive residual stress in the weld metal rather than tensile stress. Research has demonstrated fatigue life improvements of up to 100% in certain structural steel geometries using LTT consumables, with no need for post-weld treatment. The approach is currently used primarily in research and specialist fabrication; wider adoption is limited by the higher cost and narrower alloy range of available LTT fillers.
Welding Sequence and Distortion Control
Welding sequence — the order in which welds are made in a multi-weld assembly — significantly affects the accumulated residual stress field. Back-step welding reduces the total longitudinal residual stress in long weld runs by dividing the weld into shorter segments deposited in a direction opposite to the overall weld progression. Pre-setting (offsetting the joint before welding to counteract expected distortion) and balanced welding (depositing on alternating sides of a joint) reduce the overall restraint and can lower peak residual stress. These are standard procedure controls in large fabrication; see the joint types article for geometric context. For consumable selection supporting these procedures, the consumable nomenclature guide covers filler classification systems.
Fatigue Design Codes and Residual Stress Assumptions
All major structural fatigue design codes embed the effect of residual stress implicitly in their S-N design curves. Understanding this assumption is essential for correct application of code-based fatigue assessment.
IIW Fatigue Recommendations
The IIW Recommendations for Fatigue Design of Welded Joints and Components define fatigue design curves (FAT classes) based entirely on joint geometry and weld quality, with no benefit assigned to higher-strength steel. The curves are derived from large experimental datasets that include as-welded specimens with all associated residual stress. The implicit assumption is that yield-magnitude tensile residual stress is always present at the weld toe, so no mean stress correction (Goodman, Morrow) is applied for as-welded joints. For joints proven to be in a low residual stress state (through effective PWHT and measurement), IIW permits a mean stress correction, potentially allowing a higher effective stress range for the same fatigue class.
BS 7608 and EN 1993-1-9 (Eurocode 3)
BS 7608 and Eurocode 3 Part 1-9 follow the same philosophy: joint classes with associated S-N curves, no yield strength benefit, and conservative treatment of residual stress by inclusion in the experimental dataset. Both codes include provisions for improved fatigue class when the joint is subjected to predominantly compressive loading or when post-weld improvement methods are applied and documented. The mechanical testing overview covers how fatigue test data is generated and how S-N curves are statistically derived from specimen populations.
ASME Section VIII and Pressure Vessel Applications
For pressure vessels assessed under ASME Section VIII Division 2 (design by analysis), the fatigue assessment uses smooth bar S-N curves combined with a total stress concentration factor approach, rather than the weld class system of IIW. Residual stress enters the assessment through the alternating stress intensity calculation, which considers the full stress range including secondary (self-equilibrating) stresses. PWHT, where required by material group and thickness, is treated as mandatory both for toughness and for residual stress relief. The UG-84 Charpy impact testing requirements provide context for the HAZ toughness requirements that accompany PWHT specifications under ASME Section VIII.
Key References
- Watanabe O, et al. Fatigue strength of welded joint of high strength steel and its controlling factor. Q J Japan Weld Soc 1995;13:438–43.
- Ota A, et al. Fatigue strength improvement of box-welded joints using low transformation temperature welding material. Weld Int 2000;14:801–5.
- Masubuchi K. Analysis of Welded Structures: Residual Stresses, Distortion, and Their Consequences. Pergamon Press; 1980.
- Fricke W. Effects of residual stresses on the fatigue behaviour of welded steel structures. Mater Sci Eng Technol 2005;36:642–9.
- Withers PJ, Bhadeshia HKDH. Residual stress — Part 1: Measurement techniques. Mater Sci Technol 2001;17:355–65.
- Withers PJ, Bhadeshia HKDH. Residual stress — Part 2: Nature and origins. Mater Sci Technol 2001;17:366–75.
- Leggatt RH. Residual stresses in welded structures. Int J Press Vessel Pip 2008;85:144–51.
- Zerbst U. Application of fracture mechanics to welds with crack origin at the weld toe — a review. Part 2. Weld World 2020;64:151–69.
- Francis JA, Bhadeshia HKDH, Withers PJ. Welding residual stresses in ferritic power plant steels. Mater Sci Technol 2007;23:1009–20.
- Coules HE. Contemporary approaches to reducing weld induced residual stress. Mater Sci Technol 2013;29:4–18.
Recommended Reading on Welding Fracture Mechanics and Fatigue
Disclosure: WeldFabWorld participates in the Amazon Associates programme (StoreID: neha0fe8-21). If you purchase through these links, we may earn a small commission at no extra cost to you. This helps support free technical content on this site.