Welding & Welded Structural Assemblies
A weld creates both a joint and a local material and geometric condition. This article covers fusion welds, fillet welds, butt welds, resistance welding and laser welding as structural processes, addressing the heat-affected zone, weld metal, residual stress, distortion, toe geometry, root geometry, penetration, undercut, porosity, lack of fusion and misalignment — and explaining why treating a welded joint as uniform parent material in a fatigue analysis ignores the features that often govern the fatigue life.
Welding Is Both a Joining Process and a Local Thermal and Material Process
Welding is the process of joining two components by melting them together, with or without a filler material, to form a continuous metallic connection. It is the most widely used joining process for structural metallic assemblies — ships, bridges, buildings, pressure vessels, automotive structures, aerospace airframes, pipeline and Offshore structures — because it can produce a joint of high strength and high integrity without the weight and complexity of fasteners. However, welding is not simply a joining process that makes two components into one. It is also a local thermal process and a local material process: the heat of the weld melts a pool of metal, the surrounding material is heated to high temperature and rapidly cooled, and the resulting thermal cycle changes the microstructure, the properties and the residual stress state in a region around the weld. The weld also creates a local geometric condition: the weld toe (where the weld bead meets the parent material) is a stress concentration, the weld root (the underside or back of the weld) may be a stress concentration or a defect site, and the weld geometry (fillet, butt, partial penetration) determines the load transfer. A structural analysis that treats a welded joint as though the joined components simply became one uninterrupted piece of parent material ignores all of these effects — and these effects are frequently what govern the structural behaviour, particularly the fatigue behaviour, of a welded assembly.
A WELD SHOULD NOT BE MODELLED AS THOUGH THE JOINED COMPONENTS SIMPLY BECAME ONE UNINTERRUPTED PIECE OF PARENT MATERIAL.
Weld Types and Their Structural Character
The principal weld types used in structural fabrication are fusion welds (butt welds and fillet welds), resistance welds and laser welds. Each has a different structural character, a different geometric signature and a different inspection requirement. A butt weld joins two components edge-to-edge, producing a continuous joint that, if fully penetrated and properly executed, can approach the strength of the parent material in static loading. A fillet weld joins two components at an angle (a T-joint, a lap joint, a corner), producing a triangular weld bead that transfers load through shear in the weld throat; the fillet weld is the most common weld type in structural fabrication because it is simple, accessible and does not require edge preparation. A partial-penetration weld is a butt weld that does not fully penetrate the thickness; it is used where full penetration is not required or not accessible, but it has a root that is a stress concentration and a potential defect site. A resistance weld (spot weld, seam weld) joins two thin sheets by localised melting under pressure between electrodes; it is used in automotive body structures and sheet fabrication, and the weld is a discrete point or short seam that carries load in shear. A laser weld produces a narrow, deep penetration weld with a small HAZ and low heat input; it is used in precision fabrication and is increasingly used in automotive and aerospace structures for its low distortion and high speed. Each weld type has a characteristic geometry, a characteristic defect population and a characteristic fatigue behaviour, and the structural analysis must represent the weld type that is actually specified.
- Fusion weld (butt) — joins two components edge-to-edge; full penetration approaches parent material static strength; root and toe are potential critical features.
- Fusion weld (fillet) — joins two components at an angle; triangular bead; load transferred through shear in the weld throat; the most common structural weld.
- Partial-penetration weld — a butt weld that does not fully penetrate; the root is a stress concentration and a potential defect site; lower static and fatigue capability than full penetration.
- Resistance weld (spot, seam) — joins thin sheets by localised melting under pressure; discrete points or short seams; load in shear; used in automotive body structures.
- Laser weld — narrow, deep penetration, small HAZ, low heat input, low distortion; used in precision fabrication, automotive and aerospace.
The Weld Zone: Parent Material, HAZ, Weld Metal
A fusion weld creates a distinct set of material zones. The parent material is the unaffected material away from the weld; it retains its original microstructure and properties. The weld metal (fusion zone) is the material that was melted and resolidified; it has a cast structure, a chemistry that may differ from the parent (due to filler addition and dilution), and properties that depend on the consumable and the cooling rate. The heat-affected zone (HAZ) is the region of parent material that was heated to high temperature but not melted; its microstructure has been altered by the thermal cycle, and its properties may be different from — and sometimes worse than — the parent material. The HAZ can include regions of grain coarsening (near the fusion boundary, where the temperature was highest), regions of grain refinement, and regions of partial phase transformation. The width of the HAZ and the severity of the property change depend on the heat input, the material and the thickness. The transition between zones is not sharp; the properties change continuously through the HAZ. For the structural engineer, the important point is that the weld region is not uniform parent material: it is a composite of weld metal, HAZ and parent material, each with different properties, and the structural analysis must consider which zone is critical for the failure mode under consideration. For static strength, the weld metal and the HAZ may be stronger or weaker than the parent; for fatigue, the toe and the HAZ are typically critical; for fracture, the HAZ and the weld metal may have different toughness from the parent.
THE WELD CREATES BOTH A JOINT AND A LOCAL MATERIAL / GEOMETRIC CONDITION.
Weld Geometry, Residual Stress and Distortion
The diagram below shows a technical section through a fusion weld, identifying the parent material, the HAZ, the fusion zone (weld metal), the weld toe and the weld root. The toe is the point where the weld bead meets the parent material surface; it is a geometric discontinuity and a stress concentration, and it is the most common fatigue crack initiation site in welded joints. The root is the back of the weld; in a full-penetration butt weld it may be ground flush, but in a partial-penetration or fillet weld it is a stress concentration and a potential defect site. The diagram also shows the sequence: the pre-weld geometry, the thermal welding process, and the cooling that produces residual stress and distortion. The residual stress arises because the weld metal and the HAZ are heated to high temperature, expand, and are constrained by the cooler parent material; on cooling, they contract and are constrained, leaving the weld region in longitudinal residual tension and the adjacent parent material in residual compression. The distortion arises because the thermal contraction is not symmetrical: the weld is on one side of the joint, and the contraction bends the assembly. Both residual stress and distortion are inherent in the welding process and must be managed — by weld sequence, by preheat and post-heat, by stress relief, or by compensation in the design — not ignored.
[DIAGRAM: Two parts. TOP — TECHNICAL WELD SECTION: a cross-section through a fusion weld joining two plates. From left to right: Parent Material (unaffected, original microstructure) → HAZ (thermally affected, altered microstructure, shown as a shaded band) → Fusion Zone / Weld Metal (melted and resolidified, shown as the weld bead) → HAZ → Parent Material. The weld toe is marked on both sides where the bead meets the parent surface, labelled "Fatigue-critical stress concentration". The weld root is marked at the back of the weld, labelled "Potential critical feature in partial-penetration or single-sided welds". The weld penetration depth is indicated. BOTTOM — WELD SEQUENCE: three stages. (1) PRE-WELD GEOMETRY: two plates in position, gap and geometry as prepared. (2) THERMAL WELDING PROCESS: the weld torch melting the joint, heat-affected region shown as a thermal gradient. (3) COOLING → RESIDUAL STRESS + DISTORTION: the cooled assembly, with residual stress indicated (tension in the weld region, compression in the adjacent parent) and distortion shown as a slight angular bend of the assembly. A callout notes: "The weld creates a joint, a local material condition and a local geometric condition — all three must be considered."]
Weld Features and Structural Implications
The table below identifies the principal weld features, what they are, their structural consequence, their fatigue significance and the inspection approach that should be applied. The table is not exhaustive — welding is a rich process with many variables — but it covers the features that most frequently govern the structural behaviour of welded assemblies: the toe, the root, the HAZ, the weld metal, penetration, undercut, porosity, lack of fusion, misalignment, residual stress and distortion.
| Weld feature | What it is | Structural consequence | Fatigue significance | Inspection approach |
|---|---|---|---|---|
| Weld toe | The point where the weld bead meets the parent material surface | Geometric discontinuity and stress concentration; the most common fatigue crack initiation site | High — the toe geometry governs the fatigue life of most welded joints; toe grinding or re-melting can improve fatigue life | Visual and dye-penetrant for surface; the toe geometry is inspected against the specified profile |
| Weld root | The back of the weld; the underside of a butt weld or the back of a fillet | Stress concentration and potential defect site in partial-penetration or single-sided welds | High if the root is a stress concentration and is loaded; root defects can initiate fatigue cracks | Radiography for internal root; visual if accessible; the root condition must be specified and inspected |
| HAZ | Heat-affected zone; parent material thermally altered but not melted | Properties may differ from parent; can be the critical zone for strength, fatigue or fracture | Moderate to high; the HAZ microstructure affects fatigue and fracture behaviour | Hardness testing and metallography on test coupons; the HAZ is not directly inspectable in production but is controlled by the qualified procedure |
| Weld metal | The melted and resolidified material in the fusion zone | Cast structure; properties depend on consumable and cooling rate; may be stronger or weaker than parent | Moderate; the weld metal is rarely the fatigue initiation site (the toe is), but it governs static strength and fracture | Radiography for internal defects; the weld metal properties are controlled by the consumable and the qualified procedure |
| Penetration | The depth to which the weld metal fuses into the parent material | Insufficient penetration reduces the weld throat and the load-carrying capacity | High if the effective throat is reduced; partial penetration is a designed condition, not a defect, if specified | Radiography and ultrasonic; penetration is verified against the specified requirement |
| Undercut | A groove melted into the parent material at the toe, not filled by weld metal | Reduces the parent material section at the toe; a sharp stress concentration | High — undercut is a severe fatigue initiation site and is a rejectable defect in most standards | Visual and dye-penetrant; undercut is inspected against the specified acceptance criterion |
| Porosity | Gas voids in the weld metal, from gas entrapment or reactions during solidification | Reduces the effective weld section; distributed porosity is less critical than clustered or aligned porosity | Moderate; porosity is generally less critical than toe geometry for fatigue, but aligned porosity can be significant | Radiography; porosity is inspected against the specified acceptance criterion (size, distribution, alignment) |
| Lack of fusion | A region where the weld metal did not fuse to the parent or to a previous pass | A planar discontinuity that acts as a crack-like defect; severely reduces static and fatigue capability | High — lack of fusion is a severe defect and is rejectable in most standards | Ultrasonic and radiography; lack of fusion is inspected against the specified acceptance criterion |
| Misalignment | Axial or angular misalignment between the joined components at the weld | Introduces eccentricity and secondary bending at the weld; increases the local stress | High — misalignment increases the stress at the toe and can significantly reduce fatigue life | Visual and dimensional; misalignment is inspected against the specified tolerance |
| Residual stress | Locked-in stress from the thermal cycle; tension in the weld region, compression adjacent | Pre-loads the structure; can reduce effective fatigue margin; can cause distortion and stress corrosion | High — tensile residual stress at the toe can mean the fatigue crack opens under nominal compressive load | Not directly inspectable in production; controlled by the qualified procedure; may be relieved by post-weld heat treatment |
| Distortion | Geometric change from the thermal contraction of the weld | Changes the as-built geometry, the fit-up and the load path from the nominal model | Moderate; distortion increases misalignment and secondary bending, which affect fatigue | Dimensional inspection after welding; distortion is managed by weld sequence, clamp and stress relief |
Weld Fatigue: Nominal Stress, Structural Stress and Local Methods
The fatigue behaviour of a welded joint is governed by the geometry of the weld and the local material condition, not by the nominal stress in the parent material away from the weld. The weld toe is a stress concentration; the HAZ has altered properties; the residual stress is tensile at the toe. A fatigue analysis that treats the welded joint as uniform parent material and uses the nominal stress with the parent material fatigue curve ignores all of these effects and can significantly overestimate the fatigue life. The principal approaches to weld fatigue analysis are the nominal stress method, the structural stress method, the local notch method and the fracture mechanics method. Each captures a different level of the weld geometry and the local condition, and each has a domain where it is appropriate. The nominal stress method uses the nominal stress in the member and a weld classification system that assigns a fatigue category based on the weld geometry and the joint type; it is simple and widely used but relies on the correct classification of the joint and does not capture local geometry variation. The structural stress method uses the stress at the weld toe extrapolated from the structural (plate or shell) stress field, capturing the geometry of the joint but not the local notch of the toe; it is more refined than the nominal method and is used in standards for welded structures. The local notch method explicitly models the weld toe geometry and the local stress concentration; it is the most refined of the stress-based methods but requires the toe geometry to be known and modelled. The fracture mechanics method treats the weld toe as a potential crack site and assesses the growth of a crack from an initial defect; it is used for damage tolerance assessment of welded joints and for assessing the effect of known defects. This article does not invent or specify particular weld classes or fatigue curves — the applicable fatigue data and classification system depend on the standard, the material, the weld type and the application, and must be taken from the appropriate basis.
TREATING A WELDED JOINT AS UNIFORM PARENT MATERIAL IN A FATIGUE ANALYSIS IGNORES THE TOE GEOMETRY, HAZ PROPERTIES AND RESIDUAL STRESS THAT OFTEN GOVERN THE FATIGUE LIFE. The weld condition must be represented in the fatigue assessment.
Weld Analysis Approaches
The table below compares the principal weld analysis approaches. The choice of approach depends on what the analysis is trying to capture, what data is available, and what the application requires. The table is descriptive — it does not prescribe a specific approach or a specific set of fatigue data, because the applicable method and data depend on the standard, the material, the weld type and the application.
| Approach | What it captures | When appropriate | Data requirements | Limitations |
|---|---|---|---|---|
| Nominal stress | The nominal stress in the member away from the weld, classified by weld geometry into a fatigue category | Standard joint geometries with a defined classification; preliminary assessment; production joints covered by a standard | Weld classification system and associated fatigue data from the applicable standard; nominal stress from the analysis | Relies on correct joint classification; does not capture local geometry variation; not applicable to non-standard joints |
| Structural stress | The stress at the weld toe extrapolated from the structural stress field; captures the joint geometry but not the local toe notch | Standard and non-standard joints; fatigue assessment where the joint geometry is represented in the model; used in offshore, pipeline and pressure vessel standards | Structural stress at the weld toe from the FE model; fatigue data expressed in terms of structural stress | Does not capture the local notch effect of the toe; extrapolation method must be consistent with the fatigue data |
| Local notch | The local stress at the weld toe including the notch effect; explicitly models the toe geometry | Fatigue-critical joints where the toe geometry is known and the local stress concentration is significant; detailed assessment of specific joints | Weld toe geometry (radius, angle) measured or assumed; local stress from a detailed FE sub-model; fatigue data for the local notch condition | Sensitive to the assumed toe geometry; the toe radius is variable and is often a conservative assumption; requires detailed modelling |
| Fracture mechanics | The growth of a crack from an initial defect at the weld toe or root; assesses residual life and inspection intervals | Damage tolerance assessment of welded joints; assessment of known or assumed defects; inspection planning | Initial defect size (assumed or measured); crack growth data for the material and environment; stress intensity factor solution for the joint geometry | Sensitive to the assumed initial defect size; requires a valid stress intensity solution; the most data-intensive approach |
When Simple Connection Models May Still Be Adequate
The foregoing sections have emphasised that a weld is not uniform parent material and that the weld condition must be represented in the fatigue assessment. This is true, but it does not mean that every welded connection requires a detailed local model. For many structural applications, a simple connection model — in which the weld is represented as a rigid or stiff connection between the joined members, and the nominal stress in the members is checked against a classified weld fatigue category — is adequate. This is the case when the joint geometry is standard and covered by a classification system, when the load is predominantly static or low-cycle, when the fatigue spectrum is benign, or when the margin against the classified category is large. The engineering judgement is in deciding when the simple model is adequate and when a more refined model is needed. The decision depends on the joint type, the load, the fatigue spectrum, the margin, the consequence of failure and the applicable standard. A simple model that is adequate for a statically-loaded building column is not adequate for a fatigue-critical offshore tubular joint; a detailed local model that is necessary for a fracture-critical aerospace weld is unnecessary for a non-critical bracket. The principle is not "always model the weld in detail" — it is "model the weld at the level of fidelity that the application requires, and do not treat it as uniform parent material when the weld condition governs the behaviour."
MODEL THE WELD AT THE LEVEL OF FIDELITY THAT THE APPLICATION REQUIRES — BUT DO NOT TREAT IT AS UNIFORM PARENT MATERIAL WHEN THE WELD CONDITION GOVERNS THE BEHAVIOUR.