Weld toe geometry
Stress-concentrating geometrical features inherent to welding are one of the primary causes of the low fatigue strengths of welded joints compared to the unwelded material. The stress-concentrating features can be broadly classified as macro- or micro-sized; the former considers the weld bead geometry as described by the plate thickness, leg attachment length and weld toe angle and radius (Fig. 1) whilst the latter considers weld toe flaws, which typically consist of undercuts and spatter. The role of the inherent flaws in fatigue crack initiation is similar to the behaviour of sharp notches in unwelded material [2] and it has been shown that [3] even in high-quality weld flaws can have depths of 0.1 mm.
Due to the inherent inhomogeneity of welding processes, varying weld toe geometries can result, and it is currently not possible to economically identify and measure the inherent micro-size flaws [4,5,6]. Hence, a mathematically simplified or “perfect” weld toe geometry is assumed for weld toe stress and integrity assessments with the initial flaws being considered as initial cracks, with depth, a, and length 2c (Fig. 1).
To estimate stress concentration factors (SCFs) for simplified weld profiles, various parametric solutions can be found in the literature [7,8,9,10,11,12]. These are based on linear-elastic finite element modelling and are valid for specific ranges of geometrical parameters. Table 1 provides SCFs calculated for simplified weld toe geometries in the literature, with the four most recent works using a fictitious radius at the weld toe [14,15,16,17].
Table 1 SCF distribution obtained from simplified weld profiles To capture the “true” weld toe geometry, a number of methods have been used in the literature: sectioning and microscopy, laser scanning and micro-computed tomography (μ-CT) (Table 2). Nykänen et al. obtained local geometrical parameters based on published experimental data and, by using LEFM, reported their influence on the fatigue behaviour of specimens similar to the ones used in this work [29]. Some of these studies further utilised the extracted geometry and computed linear-elastic weld SCFs from it; Table 3 provides the range of SCFs obtained for each study.
Table 2 Methods used for capturing the true weld profile Table 3 SCF distribution obtained for true weld profiles In Table 3, the maximum SCF values computed for “true” weld toe profiles are higher than the values obtained from the simplified weld toe profiles. This is primarily due to more stress-concentrating notch-like geometries being captured in the former, from either undulations in the weld bead or the inherent flaws such as undercuts.
Alternating current potential drop (ACPD) fatigue crack monitoring system
The basic premise of using a potential drop method (AC or DC) for crack growth monitoring can be found in [30, 31]. Essentially PD systems measure the electric impedance of the specimen when an excitation current is passed through it. Alternating current potential drop (ACPD) measures impedance as a function of the capacitive, inductive and resistive components, whilst DCPD measures only resistance. DCPD is the more conventional method used for fatigue crack propagation growth rate measurements [30], and a comparative study of the two techniques is provided in [32].
Okumura et al. 1981 [33] used ACPD for detecting crack initiation and monitoring crack extension during stable slow crack propagation. In this study, it was assumed that crack initiation occurred at the minimum PD signal and that subsequent increases in PD were related to crack extension. Venkatasubramanian and Unvala [34] discussed the use of ACPD for crack length measurements and highlighted the impact of the positioning and joining of current and voltage probes and leads, in addition to the influence of stress on ACPD. Gibson [35] also emphasised the impact of stress on the PD signal. Tests done by Raujol-Veillé et al. [25] used a digital ACPD system developed by MATELECT Ltd. to detect the size of fatigue cracks at the weld toe of their low alloy steel non-load carrying joints. A combination of both ACPD and DCPD has been used by Wojcik et al. [36] for creep damage monitoring and end of life warning for high-temperature components.
ACPD has been used in this study due to its characteristic “skin effect” [32, 34, 36], which causes the excitation current to flow close to the surface of the specimen, as opposed to flowing uniformly through the cross-section in DCPD. The distance of the current from the surface, called skin depth, is a function of the frequency of the alternating current, with the skin depth decreasing (i.e. closer to the surface) with an increase in frequency. This effect can be used to detect small surface breaking flaws or cracks initiating at the surface; it has been used for the latter in this work. The equipment used for this test is from the same manufacturer MATELECT Ltd. as in [25, 34,35,36].
Summary
The work presented in this paper describes the development of a process that combines state-of-the-art techniques available: to non-destructively resolve the “true” weld toe profile of non-load carrying welds, produce high-resolution weld toe SCF distributions and identify crack initiation under fatigue cycling using ACPD. This approach could lead to an evaluation of the effect of the “true” weld toe geometry on the fatigue performance of a good quality manual arc-welded joint using the fatigue notch strain approach [37]. However, this is beyond the scope of this paper and will be discussed in a future publication.