Facing multiaxial fatigue testing challenges with respect to non-proportional loading conditions, a custom-built hexapod has been used to establish the mode-{I, III} resistance characteristics of high-quality welds in steel maritime structures. Assessment of the hexapod test data using the effective notch stress and total stress, respectively the best performing multiaxial intact and cracked geometry parameters, shows a fit in the reference quality literature data scatter band and provides conservative lifetime estimates. In order to improve the lifetime estimate accuracy, strength, geometry, material and mechanism aspects are investigated. Welding induced residual stress, a strength aspect, predominantly affects the mode-I fatigue resistance including a mean (residual) stress contribution. The weld notch radius, a geometry parameter, primarily influences the mode-III fatigue resistance. Similar material microstructure compositions of the high-quality welds and reference quality ones are observed, implying comparable mode specific mechanism parameters for the effective notch stress and total stress, respectively the material characteristic length and elastoplasticity coefficient. The material microstructure properties and classification criteria for high-quality welds support the residual stress estimates and suggest a smaller welding induced defect size. In general, the high quality is mainly reflected in the larger resistance curve intercept and slope, another strength and mechanism parameter, implying a larger initiation contribution to the total lifetime. For a high-quality resistance curve involving the representative strength, geometry, material and mechanism contributions, more accurate lifetime estimates are obtained, even though the parameter confidence is reduced because of the relatively small data size in comparison to the reference quality one.

Maritime structures, such as offshore support vessels and floating wind turbines, face time-varying loads from environmental elements like wind and waves, as well as operational loads from machinery. This exposure leads to fatigue, a progressive cracking process triggered by cyclic loading, which is a significant concern for structural integrity. Fatigue cracks often originate at stress concentration points, which can occur at different scales due to material defects and structural features. In steel structures, welded joints are particularly vulnerable due to their stress-concentrated geometry, making fatigue analysis crucial for their longevity.

Fatigue in maritime structures is inherently multiaxial due to various external loads and complex structural responses, involving mode-I, mode-II and mode-III loading. In typical maritime applications, mode-I induced damage is inherently governing, but multiaxial loads from torsional moments and out-of-plane forces can also impact fatigue life. Modelling these multiaxial loads accurately is necessary to predict fatigue life in welded joints, which experience both fatigue initiation and crack growth phases.

Fatigue life modelling involves assessing both intact and cracked geometries. For welded joints, fatigue life is typically crack-growth dominated due to inherent defects from welding. Linear elastic models, such as Basquin’s equation, often relate stress to fatigue life, showing a log-log linear relationship. Mode-I and mode-III loading modes differ in their strength and mechanism contributions, which can vary based on whether the load cycles are proportional or non-proportional.

The research focuses on a systematic approach for multiaxial fatigue analysis of arc-welded joints in steel maritime structures. The goal is to develop an accurate, reliable, and simple methodology for assessing fatigue life. This includes selecting appropriate fatigue criteria and developing robust cycle-counting techniques for handling non-proportional multiaxial loads. A von Mises stress-based failure criterion is used, supported by a mode-dependent shear strength coefficient to improve the accuracy of fatigue predictions. Literature-derived coefficients help capture the distinct behaviour of mode-I and mode-III contributions.

To evaluate mode-III response characteristics, new formulations for weld toe notch stress distributions were developed. A linear damage accumulation model offers simplicity in life estimation, while semi-analytical methods improve prediction accuracy. Both intact and cracked geometry parameters show reliable life estimates, though each brings specific strengths – the former for averaging capability and the latter for detailed physics of fatigue.

Fatigue resistance data available in the literature were used to create a general S-N design curve for typical weld quality. For high-quality welds, new fatigue tests conducted using a specially designed hexapod platform revealed higher resistance than standard predictions. This high-capacity, six-degree-of-freedom testing apparatus allows rigorous multiaxial fatigue assessments, filling gaps in non-proportional load data and providing a foundation for improved life predictions.

Ultimately, by combining theoretical modelling, experimental data, and novel testing methods, this research advances multiaxial fatigue assessment in maritime steel structures, with findings that emphasize the critical need for mode-specific contributions and the potential benefits of enhanced testing for accurate fatigue life predictions.

The response of maritime structures can be multiaxial, involving predominant mode-I and non-negligible mode-III components. Adopting a stress distribution formulation based effective notch stress as fatigue strength parameter for mixed mode-{I, III} multiaxial fatigue assessment purposes, a mode-I equivalent von Mises type of failure criterion has been established at the critical fracture plane. Counting includes a cycle-by-cycle non-proportionality measure and damage accumulation is based on a linear model. Distinguished mode specific and material characteristic strength and mechanism contributions in terms of respectively the resistance curve intercept and mean stress induced response ratio coefficient, resistance curve slope and material characteristic length, have been incorporated. Evaluating the mid-cycle fatigue resistance, the outperformance is impressive. The analysed multiaxial mode-{I, III} data fits the uniaxial mode-I reference data scatter band and a single resistance curve can be used for fatigue assessment.

Arc-welded joints in steel maritime structures are typically identified as weakest links in terms of fatigue limit state performance. Multiaxiality can be involved, consisting of predominant mode-I and non-negligible mode-III components. Aiming to answer the question if a cracked geometry based fatigue strength parameter would outperform an intact geometry based one like the effective notch stress, the total stress is adopted. A von Mises type of criterion is defined at the critical fracture plane and includes mode specific and material characteristic strength and mechanism contributions. A lifetime dependent shear strength coefficient is introduced to cover the resistance curves intercepts and slopes, whereas the total stress parameter contains the mean stress contribution as well as the (mixed) mode dependent notch and crack tip elastoplasticity coefficients, reflecting an interaction mechanism. Cycle counting includes a cycle-by-cycle non-proportionality measure and damage accumulation is based on a linear model. Evaluating mid-cycle fatigue resistance data, the total stress and effective notch stress performance turns out to be similar. However, the total stress related elastoplasticity coefficients are an explicit and sensitive measure to incorporate the actual physics of the fatigue damage process, whereas the material characteristic lengths for the effective notch stress seem to be more implicit and less sensitive ones.

The predominant mode-I response of maritime structures can be multiaxial, involving out-of-plane mode-III shear components. Semi-analytical mode-III notch stress distribution formulations have been established for critical details like welded T-joints and cruciform joints, reflecting (non-)symmetry with respect to half the plate thickness. Using a stress distribution formulation based effective notch stress as fatigue strength criterion, the mode-III welded joint mid-cycle fatigue resistance characteristics have been investigated. In comparison to mode-I, the material characteristic length and resistance curve slope estimate suggest the fatigue damage process to be even more an initiation related near-surface phenomenon. Mean shear stress effects seem insignificant.