PSI - Issue 52

Hongmin Zhu et al. / Procedia Structural Integrity 52 (2024) 679–689 Zhu et al./ Structural Integrity Procedia 00 (2019) 000 – 000 5 Furthermore, the localisation error is measured for comparing the damage localisation accuracy of the 3 baseline free techniques formulated as = √( − ) 2 +( − ) 2 (11) where ( , ) and ( , ) correspond to the predicted and the true damage locations. Note that the localisation error is primarily utilised for single damages. 683

Fig. 1. The 400 × 400 CFRP plate with simulated damages.

4. Results The 12-sensors CFRP coupon is firstly investigated under 20℃ within 10 single damage scenarios where each single damage has the radius of 15 and the imaging results are obtained. Among the results four typical scenarios are illustrated in Fig. 2. It is shown that the IB, VTR and the reciprocity-based methods are able to detect and localise single damages at different locations within the circular transducer array. Among the three methods, the VTR and the reciprocity-based techniques yield similar overall accuracies and variances, which are superior to the IB method, when localising the four damages. Furthermore, the simulated damages from different locations within the circular transducer network can also be well detected and localised by the VTR technique and the reciprocity principle due to high sensor coverage at the locations. Comparatively, the IB technique yields modest localisation performances and may even fail for accurate damage localisation under specific damage scenarios. For example, when the potential damage is located at the centre of the 12 transducers, in the case of the damage 4, such damage cannot be accurately localised by the IB technique considering that no equivalent through-centre transducer paths can be utilised for the anisotropic CFRP plate. Moreover, overall localisation performances of the three baseline-free techniques under the 10 single damage scenarios are illustrated and compared in the Fig. 3. It is shown that the VTR technique yields the lowest localisation errors as 8.6 and error variations as 1.1 , which indicates high effectiveness and robustness of such technique in terms of the single damage scenarios. Meanwhile, the reciprocity-based technique has a moderate performance among the compared methods with the localisation error and the error variation as 11.1 and 1.7 respectively. The IB technique, however, leads to the poorest performance when localising the 10 single damages mainly because such technique is highly dependent on sensor positionings and can be held back by specific damage scenarios, e.g., the damage case 4 where a putty damage is simulated at centre of the circular transducer arrays and no equivalent paths can be utilised considering the anisotropy of composite structure. Similarly, Fig. 4 illustrates imaging results of single damages on the 12-sensors coupon under 20℃ , 30℃ and 40℃ . It is shown that the investigated baseline-free techniques are capable of identifying and localising damages under temperature variations of 20℃ .