4. Discussion
Under quasi-static loading, the knowledge of local temperature changes helps understanding the thermal examination through the bonded patch. Monitoring a patched specimen, sudden local temperature increases at an angle of around 45° to the loading direction, combined with noise effects in the load/displacement curve can thus be allocated to the PLC effect. An increase of the effect over time is measurable. Compared to the specimens without patch the detectable temperature difference from the patched side is reduced, because temperature differences are rather small, and the resulting infrared radiation is weakened by the CFRP patch. Additionally, temperature flashes run suddenly through the specimen so that a clear identification of damage initiation through the patch is difficult. Further, due to the local patch detachment the air between the patch and the specimen additionally influences the radiation intensity. For patches of one layer of the chosen CFRP material, subsurface effects which are related to heat generation can generally be identified on thermal images. But, temperature difference \(T\) has to be high enough. The origin of heat generation cannot clearly be identified without further information. Interpretation of thermal images needs for additional knowledge resulting from the stress/strain behaviour e.g.
For specimens loaded in fatigue, results are different. Crack length measurement is possible from the metallic side and becomes even better, the longer the crack growth. As the specimens are loaded by the lower piston, they heat up from the bottom to the top. The temperature difference between the upper and the lower part of the specimen increases with increasing number of load cycles. As the heat circles the crack the crack itself acts as a temperature barrier and becomes visible through the temperature difference of the metallic part above and the one below the crack. From the patched side it is not as clear, but a good qualitative estimation is possible as well. Further, under cyclic loading, crack growth additionally goes along with local patch detachment, leading to a further reduction of crack detectability. A comparison of the results monitored from the metallic and from the patched side shows that crack length detection is possible from both sides. As seen for the unpatched specimen, a discontinuous temperature distribution appears around the crack tips with a maximum temperature increase \(T\) at the peak stress of each loading/unloading cycle. Using this top dead centre for test evaluation, crack tips can be identified also from the patched side. Imaging frequency should therefore be higher than test frequency so that it can be made sure that pictures at or close to the top dead centre are available. Unlike for the quasi-static loading, here, the measured temperature of the metallic and the patched side do not differ strongly, as long as there is a proper bonding between patch and specimen. Increasing local damages in the adhesive layer have an impact on the thermal properties resulting in decreasing visibility of the crack. Local patch delamination can be identified on the thermal image as areas where the woven structure of the patch is noticeable. But still, crack propagation process can be monitored. Comparing crack length and detachment dimensions it seems that right in front of the crack the patch has already delaminated from the metal. Meaning that for the given specimen configuration and the given conditions presumably delamination around the crack tip appears before the crack grows. To further corroborate this hypothesis additional tests using complementary monitoring techniques have to be made.