PSI - Issue 77

João Nunes et al. / Procedia Structural Integrity 77 (2026) 593–600 Joa˜o Nunes et al / Structural Integrity Procedia 00 (2026) 000–000

598

6

The two bolt positions after induced loosening can be clearly distinguished in periods 2 and 3, as shown in Figure 5. A Pearson correlation coe ffi cient of 0.96 was achieved, indicating that both systems—DIC and strain gauges—reliably detected the induced loosening of the tightening system. In turn, the magnitude di ff erence between the strain-gauge and DIC results stems from the experimental setup. The DIC monitored the intentionally loosened shaft, therefore reporting higher strain values, whereas the strain gauge was placed on a di ff erent shaft where no loosening was induced as demonstrated in 2 b) . Consistent with the results discussed in Section 3.1, both shafts once again experienced tensile loading due to membrane expansion, as expected. 3.2.1. Thermography To evaluate the feasibility of using thermographic cameras for temperature monitoring, the thermal images were converted to MATLAB files using FLIR ResearchIR software and subsequently compared with the thermocouple measurements. During the analysis, a region of interest (RoI) was selected in the captured images, Figure 6 a), which corresponded approximately to the shaft region where the thermocouple was placed. The results obtained with both methods are presented in Figure 6 b).

Fig. 6. a) Thermographic image of the hydrogen cell and Region of Interest, and b) Comparison between thermocouple and RoI mean temperature

The thermographic camera’s results showed a slight increase in temperature, followed by a decrease when the cool ing fan is activated. The consistency observed between the thermocouples’ and the thermographic camera’s results, with a Pearson correlation coe ffi cient of 0.82, demonstrates that thermography is a suitable non-destructive inspection technique for assessing heat variations in a hydrogen fuel cell. The di ff erences in magnitude observed in Figure 6 b) can be attributed to the fact that the thermographic setup results represent the mean over a larger RoI.

4. Conclusions

The present study allowed for a successful understanding of the overall hydrogen cell behaviour during short-term operation. By employing thermocouples and a strain gauge, it was confirmed that the tightening system shafts are subjected to tensile loading due to membrane expansion during cell operation. Simultaneously, the tightening system’s shafts were validated as a suitable location for the future integration of printed sensors for structural monitoring since the cell´s tightening system loosening was successfully identified in these components. Additionally, it was concluded that, in this cell model, the cooling fan is activated a few minutes after cell start-up to counteract the initial increase in temperature. DIC was evaluated against strain gauge measurements positioned on a di ff erent shaft. A deliberate loosening of the tightening system was performed to simulate operating conditions. The comparison determined a high Pearson correlation coe ffi cient (0.96), indicating a strong correlation in detecting loosening bolts. Although DIC and strain gauge signals showed similar temporal trends, the DIC-measured strain magnitudes were higher as the DIC monitored the loosened shaft, while the strain gauge was placed on a di ff erent shaft. Similarly, the thermographic setup was successfully validated against thermocouples with a Pearson correlation coe ffi cient of 0.80. Finally, the experimental setup was prepared to acquire the hydrogen cell output voltage and current, enabling the development of a comprehensive dataset integrating performance and structural integrity metrics.