PSI - Issue 39

Hannes Panwitt et al. / Procedia Structural Integrity 39 (2022) 20–33 Author name / Structural Integrity Procedia 00 (2019) 000–000

21

2

Keywords: 3D mixed mode; Crack length measurement; Crack branching

1. Introduction Since machines and plants are often exposed to a very complex and also variable load situation in practice, the three basic crack modes I, II and III do not usually occur individually at a fatigue crack. Instead, the crack modes are often superimposed to a mixed-mode loading. Today's studies in this field are mainly concerned with the plane mode I/II combination or a mode I/III loading (e.g., by Fatemi et al. (2014), Fonte et al. (2015), Eberlein and Richard (2016) and Yaren et al. (2019). However, the simultaneous superposition of mode I, II and III is investigated much less frequently. In addition, the corresponding test conditions are currently not standardized, so that different specimens and test methods are mostly used, which sometimes lead to contradictory results (Socie and Marquis, 2000). For these reasons, there are still deficiencies in the theoretical description of the crack propagation behavior taking into account all three crack modes (Köster et al., 2020). This is again very critical with regard to the prediction of component service life and the evaluation of safety in operation. To evaluate the crack propagation behavior, it is important to know the exact crack path and the position of the crack tip. Common examination methods are, for example, the direct current potential drop technique or the use of strain gauges. Their application works very well especially for the very straight and plane mode I cracks. However, mixed mode loading usually produces very complex, non-planar crack geometries, so that the established methods of crack length measurement are not sufficient in this case. Instead, the use of optical measurement techniques is more appropriate (Khalili and Vahidnia, 2015; Simunek et al. 2018). In particular, the digital image correlation (DIC) method, which has been used more and more frequently in recent years (Neggers et al., 2018), provides a good basis for such applications. However, due to the very different crack paths, the inherent noise of the DIC measurement (Zhao et al., 2019) and the high number of images to be evaluated, the application of this method is associated with a very high manual effort (Köster et al., 2020). To overcome the limitations of existing crack detection methods Breitbarth et al. (2021) developed an automated evaluation of the DIC data. They trained a convolutional neutral network for crack path detection and further calculated the crack tip load in terms of the stress intensity factors K I and K II based on the interaction integral. Furthermore, Gehri et al. (2020) proposed a new concept for an automated crack detection and measurement (ACDM), which was primary developed to detect multiple cracks in brittle materials such as concrete under quasi static loading and moreover to analyze crack kinematics such as slip and width. This tool also delivers a good basis for crack investigation in metallic materials under cyclic loading, but does not provide information about the crack growth in terms of the a-N- curve. Also, a method to deal with a branching crack is needed for the evaluation of mixed mode fatigue cracks. For this purpose, in the following an extension of the ACDM-script is proposed to address these limitations and enable an automated crack detection and crack growth measurement for mixed mode fatigue crack growth. In addition, the experimental studies conducted to calibrate and validate the advanced features of the tool are presented.

Nomenclature a

crack length

crack length increment

∆ a

branch number crack number

b c

elongation at fracture Young’s modulus

A 5

E

maximum force

F max

force for single overload

F ol

stress intensity factor for mode I stress intensity factor for mode II

K I K II