PSI - Issue 66

Yamato Abiru et al. / Procedia Structural Integrity 66 (2024) 525–534 Author name / Structural Integrity Procedia 00 (2025) 000–000

532

8

Crack-free

N =70,000 ( a = 0.78 mm)

N =120,000 ( a = 2.04 mm)

N =230,000 ( a = 3.26 mm)

N =450,000 ( a = 4.0 mm: Reached outer surface of pipe)

Fig. 9. Results of hammering test for both ends supported beams of hydrogen-precharged pipe specimen. Refer to Fig. for striking points A-C and to Fig. 5 for corresponding crack lengths in thickness direction.

N = 13,121 from Fig. 5, which may underestimate the actual crack length. These results highlight the need for improved detection accuracy while acknowledging that hydrogen-precharged specimens may exhibit slightly reduced sensitivity compared to uncharged specimens. Natural frequencies of up to 10 kHz are observed in pipeline systems, and the results indicate that crack conditions influence characteristics within lower frequency ranges. For larger hydrogen infrastructure components, such as tanks or valves, the natural frequencies increase due to their larger volume. Therefore, combining crack-detection methods across higher frequency ranges is recommended, tailored to the specific dimensions and shapes of the components. However, hydrogen-induced cracks often propagate along straight paths, making manual inspection challenging and infrequent, which underscores the need for automation. Inspecting an entire pipeline remains difficult. Besides monitoring crack propagation, inspections must also encompass stress concentration and thread loosening at pipeline joints. Therefore, conducting a stress analysis to identify weak points and limit inspection to critical areas is essential. The use of deep-learning technology could further enable automation and improve inspection precision, ultimately enhancing system safety. Advancing detection accuracy, achieving full automation, as well as establishing supportive legislation remain key challenges in this area.