PSI - Issue 66

Hendrik Baarssen et al. / Procedia Structural Integrity 66 (2024) 305–312 Author name / Structural Integrity Procedia 00 (2025) 000–000

311 7

(a)

(b)

Figure 4. a) The measured area captured by one of the cameras b) the location of the crack as determined by the algorithm

Figure 5 (a) shows the crack length as determined by the algorithm with different threshold values as a function of the number of acquired images. It is clear that a threshold of the mean of the noise results in unreliable crack detection, but the results are significantly improved when the threshold is increased with 1 standard deviation of the noise and converges for the higher considered threshold values. Figure 5(b) shows a comparison of the detected fatigue crack growth using the DIC, the crack gauge, and the CMOD compliance. Up to image 32, the crack length measured by the three methods is practically overlapping. In the last images, a small deviation is observed for the crack length estimated by the CMOD compliance, potentially due to plasticity and crack tunneling. The DIC and crack gauge still show comparable results.

(a)

(b)

Figure 5. a) fatigue crack growth for different threshold values b) comparison of the fatigue crack growth measured with DIC, crack gauge and CMOD compliance 4. Conclusion In this paper, a framework to detect the location of the crack tip based on displacement fields measured by digital image correlation (DIC) and to measure the crack length is proposed, hence enabling the measurement of the fatigue crack growth rate. The framework is applied in an experimental investigation involving a C(T) specimen, and the results are compared with the measurements using a crack gauge and the CMOD compliance method. The results show that the crack can be measured using the proposed framework over virtual line gauges corresponding to column vectors of the matrix storing the displacement field. A threshold for the Chow coefficient is calibrated based on the Chow coefficient determined before crack initiation. Since the threshold can be calibrated based on DIC data,