The main objective of the current research is to better understand the combined effects of weld-induced microstructures, residual stresses and stress concentrations on the fatigue behavior of welds subjected to complex multiaxial cyclic loadings. As such, characterization of welding residual stresses, relaxation and redistribution of residual stresses under mechanical loading, microstructure-dependent material deformation behavior, fatigue damage quantification and notch effect were the key areas of investigation. Comprehensive experiments including deformation tests on round solid and tubular samples, and fatigue tests on tubular weld specimens made of S355J2H steel were conducted to support the analysis. To assess the role of residual stresses in fatigue damage, two groups of similar weld samples, with and without residual stresses, were compared.
In weld integrity assessment, accurately capturing the interaction between load and residual stresses is crucial. Traditionally, this has involved simplifications, such as assuming uniform residual stress distribution. While advanced experimental and numerical methods now allow for more precise determination of residual stress magnitude and distribution, integrating all influencing thermal, metallurgical, and mechanical factors remains challenging.
... mehr
The reliability of predicted residual stresses strongly depends on the assumptions and simplifications made. Given the limitations of both experimental and numerical methods, a combined approach is essential. Recent software advances have improved computational weld mechanics, but a deep understanding of the underlying physics and theory remains crucial. This research addresses key welding phenomena, including temperature-dependent deformation and metallurgical phase transformations.
Residual stress relaxation can occur partially or fully under thermal or mechanical loads if the combined stresses exceed the material's yield strength. Beyond their magnitude and distribution, the relaxation and redistribution of residual stresses under operational loads are critical for accurate weld design and life prediction. If significant residual stresses remain after relaxation, they may combine with load stresses, affecting fatigue life. Ignoring relaxation can lead to inaccurate life estimates, reducing reliability and leading to overdesigned, cost-inefficient structures. Relaxation behavior varies with material type and loading conditions.
Numerous methods exist for fatigue assessment and lifetime estimation of components under multiaxial stress states. Some models extend classical static material strength hypotheses, such as equivalent stress/strain quantities, to multiaxial conditions, assuming an equivalent uniaxial stress; these invariant approaches yield acceptable results under proportional or In-Phase (IP) loading but often overestimate fatigue life under non-proportional or Out-of-Phase (OP) conditions due to their inability to capture non-proportional hardening. Advanced analysis methods, offering insight into the physical background of fatigue failure under uniaxial and multiaxial loading, are therefore emphasized in this study. Another group of models calculates a damage parameter on specific planes where fatigue cracks initiate, with critical plane approaches focusing on a single preferred plane and being classified as stress, strain, or energy-based models. These approaches are well-suited for multiaxial stress states arising from external loading or inherent features like notches or weld toes, and can capture both Low Cycle Fatigue (LCF) and High Cycle Fatigue (HCF), including non-proportional loading effects.
In this research, the strain-based Fatemi-Socie (FS) damage parameter was used to quantify fatigue damage, providing improved life estimations over conventional methods and effectively correlating multiaxial fully reversed constant amplitude fatigue data of heat-treated (residual stress free) specimens. For as-welded samples with residual stresses, the modified FS parameter offered even higher accuracy in correlating data and predicting fatigue life.
To account for the notch effect on local deformation and stress gradient, a more robust approach called Theory of Critical Distance (TCD) was implemented together with a constitutive plasticity model which leads to a better correlation of multiaxial fatigue data in comparison to the conventional approaches such as the Neuber’s rule with a fatigue notch factor.
Due to significant microstructural phase transformations caused by welding in the weld area and Heat Affected Zone (HAZ), a detailed study compared the material behavior of the Base Material (BM) and HAZ under various cyclic loading conditions. The HAZ microstructure near the weld was reproduced via specific heat treatment, followed by deformation tests on both BM and HAZ. Material characterization revealed notable differences between the base material (BM) and heat affected zone (HAZ) under monotonic and cyclic loading. Concurrently, the cyclic Chaboche plasticity model was implemented in a Finite Element (FE) model, with parameters calibrated using complex multiaxial tests. To incorporate Non-Proportional Hardening (NPH), Tanaka’s constitutive model was coupled with the Chaboche model by modifying the isotropic hardening term.
In addition, both materials exhibited cyclic transient softening within a specific strain range. To account for this behavior, a damage-dependent cyclic stress-strain curve was incorporated into the lifetime assessments alongside the modified FS damage model. Since the FS damage parameter includes both stress and strain components, it effectively captures material hardening (or softening) effects in fatigue life predictions. This approach to modeling cyclic transient softening resulted in improved accuracy of life predictions in the LCF regime.
While this study focused on fatigue crack initiation in welded joints, it also briefly examined micro/macro crack propagation using a discrete crack growth approach. This method effectively predicts both crack initiation and propagation, implicitly accounting for the notch effect in a more realistic way. The greatest improvement in fatigue life prediction was observed in pure axial tests, while pure torsion tests showed minimal impact due to a weaker stress/strain gradient at the notch—making crack propagation life negligible compared to initiation. The approach also captures microstructural influences on cracking through both plasticity behavior and cracking mechanisms.
