Creep in Welded Structures
Creep in welded structures — the weld metal, the heat-affected zone and the parent material, property mismatch across the weld, local stress concentrations, creep damage localisation and the role of residual stress.
The welded structure creep problem
Welded structures at elevated temperature present a unique creep challenge: the weld is not a single material but a composite of weld metal, heat-affected zone (HAZ) and parent material, each with different microstructures and different creep properties. The property mismatch across the weld creates local stress concentrations and causes the creep damage to localise in the weakest region. The weld may be the life-limiting feature of a high-temperature structure — many service failures of high-temperature piping and pressure vessels occur at welds. The assessment must model the weld as a composite, with the different regions and their respective properties, and must identify the region where the creep damage localises. A simplified analysis that treats the weld as a single material with parent-metal properties will miss the local damage and may be non-conservative.
Weld metal, HAZ and parent material
A weld consists of three distinct regions. The weld metal is the deposited material (the filler), which has a cast microstructure that differs from the wrought parent material. The heat-affected zone (HAZ) is the parent material that has been thermally affected by the welding heat — the microstructure has been modified (austenitised and re-transformed) but the material has not been melted. The parent material is the unaffected base metal. Each region has different creep properties. The weld metal may be stronger or weaker than the parent, depending on the filler selection. The HAZ may be the weakest region — in some alloy systems (e.g. some Cr-Mo steels), the HAZ has a significantly lower creep strength than the parent, due to the microstructural change (e.g. carbide coarsening, grain refinement). This is the basis of Type IV cracking, a common failure mode in high-temperature welds.
Property mismatch and local stress
The different creep properties of the weld regions create a property mismatch. Under load, the regions creep at different rates. The weaker region (e.g. the HAZ) creeps faster, but it is constrained by the stronger adjacent regions. The constraint produces local stress concentrations at the interfaces between the regions. The stress concentration may be significant — the multiaxial stress state at the interface can be much higher than the nominal stress. The local stress concentration, combined with the weaker material, causes the creep damage to localise in the weakest region. The damage localisation is the key feature of weld creep — the damage is not uniformly distributed but concentrated at a specific location (typically the HAZ, or the interface between the HAZ and the parent). The assessment must model the property mismatch and capture the local stress concentration.
Creep damage localisation — Type IV cracking
In some alloy systems (notably some Cr-Mo and Cr-Mo-V steels), the creep damage localises in the intercritical region of the HAZ — the region that was heated to just below the Ac3 temperature during welding. This localised damage is called Type IV cracking. The intercritical HAZ has a refined grain structure with coarsened carbides, giving it a lower creep strength than the parent material. The damage manifests as creep cavities and microcracks in the intercritical HAZ, which link to form a macrocrack that propagates around the weld. Type IV cracking is a major concern in high-temperature piping and pressure vessel welds. The assessment of Type IV cracking requires a model that represents the intercritical HAZ properties (which are lower than the parent) and the local stress state. The assessment is typically conservative if the parent material properties are used (the actual HAZ is weaker), but the degree of conservatism (or non-conservatism) should be evaluated.
Residual stress and its relaxation
Welding residual stresses are present in the as-welded structure. The residual stress may be up to the yield strength in the weld and the HAZ. At elevated temperature, the residual stress relaxes by creep. The relaxation is beneficial (it reduces the tensile residual stress) but it accumulates creep strain and damage. The combined effect of the residual stress and the applied stress determines the initial stress state for the creep analysis. The residual stress should be included in the initial state if it is significant. In some cases, a post-weld heat treatment (PWHT) is performed to relieve the residual stress before service — if the PWHT is effective, the residual stress may be neglected. If there is no PWHT, or if the PWHT is partial, the residual stress should be included. The relaxation of the residual stress during the early service life may cause significant creep damage in the weld and the HAZ.
Assessment of welded structures
The assessment of a welded structure for creep should model the weld as a composite with the weld metal, the HAZ and the parent material, each with their respective creep properties. The properties should be obtained for the specific welding procedure (filler, process, post-weld heat treatment) — the creep properties of welds are procedure-specific and cannot be assumed from the parent material properties alone. The FEA should include the weld geometry (the weld profile, the HAZ width, the root and the cap) and the property variation across the weld. The assessment should identify the critical location (which may be the HAZ, the weld metal or the parent, depending on the relative properties) and predict the creep strain, the damage and the rupture life at that location. The assessment should also consider the possible failure modes — Type IV cracking, weld metal rupture, or parent material rupture — and ensure that the analysis captures the relevant mode. For safety-critical welds, the assessment should be supported by weld-specific creep test data (cross-weld creep tests, or tests on simulated HAZ material).
A weld is not a single material — it is a composite of weld metal, HAZ and parent material with different creep properties. The property mismatch causes stress concentrations and damage localisation, often in the HAZ. Always model the weld as a composite with region-specific properties. Treating the weld as parent material will miss the local damage and may be significantly non-conservative.
Weld-zone representation in analysis
A welded high-temperature structure should not automatically be represented as homogeneous parent material. The appropriate level of detail depends on the failure mechanism and the available data. For global deformation, an effective homogeneous representation may be sufficient; for local creep damage or Type IV assessment, separate parent, weld metal and heat-affected-zone regions may be necessary. The HAZ width in an FE model should be based on a defensible metallurgical or test basis rather than on mesh convenience, because an artificially narrow weak zone can create mesh-sensitive localisation. Material axes, weld geometry, mismatch and local temperature should also be represented where they influence the result. Sensitivity to HAZ width and property set is often a more meaningful robustness check than simply refining the mesh. The modelling resolution should be matched to the evidence available: a highly detailed zoned model is not automatically more credible if its zone properties are poorly characterised.
Linking analysis to inspection and crack evidence
For service-exposed welds, creep assessment should be integrated with inspection findings and known damage morphology. Surface and volumetric inspection, replication, hardness measurements and metallographic evidence can help distinguish distributed creep deformation from localised cracking. If cracking is already present, a continuum creep analysis alone may no longer answer the integrity question and a fracture or crack-growth assessment may be required. The FE model can still be valuable for identifying the stress and strain field driving the observed damage and for prioritising similar joints. A strong remaining-life argument therefore states clearly whether the structure is being assessed as uncracked, damage-tolerant or already cracked, and aligns the analytical method with that state rather than applying a generic creep-damage metric to all three.