PSI - Issue 57

Kalle Lipiäinen et al. / Procedia Structural Integrity 57 (2024) 32–41 Lipiäinen et al. / Structural Integrity Procedia 00 (2019) 000 – 000

33

2

1. Introduction Additive manufacturing(AM) of metallic materials is fast developing method for high-performance applications. AM includes multiple techniques like laser-based direct energy deposition (L-DED), laser-based powder bead fusion (L-PBF), wire arc additive manufacturing (WAAM). The methods have different characteristics in terms of dimensional accuracy, porosity measured as material density, surface roughness and initial crack-like defect size. Solberg (2021) presented design and verification procedure considering fatigue properties for PBF-components. Hensel et al. (2022) studied WAAM specimens fatigue performance with local effects considered. AM-components are typically optimized in digital production chain. It is usually straightforward to optimize tensile strength. However, fatigue strength assessment requires more complex approach than maximum use or overload cases. Artificial intelligence (AI) and machine learning (ML) can be used to evaluate components experiencing fatigue damage. Wang et al. (2022) analyzed importance of input features for fatigue strength assessment and naturally found external stress to have the most important influence on fatigue performance followed by surface roughness and compressive residual stresses on surface. The benefit of AI is very limited when only traditional values like surface roughness and material properties, which are easy to define numerical values, are used as an input. Li et al. (2022) used computervision and ML, to evaluate defects influence on fatigue performance via L-PBF 17-4 PH specimens fractography. In this study, fractography is converted to a numeralvalue for additional parameter which is not typically included in component fatigue strength assessment. The internal crack-like defect length is in many cases considered only with outside surface roughness and correction factors to modify surface roughness. Lipiäinen, Afkhami, et al. (2022) used surface roughness as a basis for analysis in but found later that the study lack of not so obvious internal quality parameterwhich was introduced for constant amplitude (CA) (Lipiäinen, Ahola, Kaijalainen, et al., 2022) and for variable amplitude (VA) loading (Lipiäinen, Ahola, & Björk, 2022). This study highlights the quality and local effects on AM-components ’ fatigue strength assessment. Fatigue strength analysis based on √ extracted from each specimen with SEM fractography explains performance differences between test series. This study aims to introduce an engineering method that is simple to utilize from 2∙10 4 cycles to run out -level around 10 6 cycles and could be extended on VA-load. Authors previous studies specimens are re-evaluated with novel method. The multiparametric theory of critical distances (TCD) -based concept if further clarified with illustrated examples of differentiating nominal, geometric and quality-based stress components and their material matched cyclic behavior. 2. Multiparametric 4R method 4R method is a parametric local cyclic behaviour-based fatigue strength assessment tool that has been developed by Nykänen & Björk (2015) and further by Ahola et al. (2020) . 4R method features graphicalinterpretation of local cyclic behaviour. Schematic local cycle behaviour illustrated in Figure 1 shows the effect of SCFs (local quality or geometry) and material strength in respect to local stress ratio. To utilize 4R method, it is crucial to differentiate nominalstress (red line), geometrical stress (blue line) and quality induced stress (black line). CX specimens provide interesting setting as the specimens are tested in as-built and heat-treated conditions with major difference in material strength. Heat treated specimens feature higher local stress ratio R due to elongated linear regime. Stress components classification for nominal, geometric and quality- based stresses in Fig 1 is used for visual interpretation only. The stress components used in fatigue strength assessment should match true-life to enable evaluation without additional adjustment factors by obtaining physical data from specimens or structures. An accurate relationship between stress obtaining method and material model is needed for the reliable utilization of the parametric method. Especially on case of variable amplitude loading, the local stress ratio has an important role and the accuracy of fatigue strength assessment significantly decreases if local properties have been defined in a incorrect way. To model sharp crack-like defects, theory of critical distances (TCD) -method by Taylor (2008) was implemented by the authors to extract local stresses for the multiparametric 4R method.