PSI - Issue 52

Ilias N. Giannakeas et al. / Procedia Structural Integrity 52 (2024) 655–666 Author name / Structural Integrity Procedia 00 (2019) 000 – 000 9 The process is repeated for all impact events and temperature difference values. At each one, the mean ( ̃ ) and standard deviation ( ̃ ) of the posterior of ̃ are computed. These values correspond to the accuracy and precision of the predictions. The ̃ and ̃ values for each impact event are plotted in Fig. 7. At low values, both ̃ and ̃ are close the values that were obtained when perfect information has used (see Fig. 4). This ca be explained because at = 0 , is close to zero and is small. However, as the temperature difference increases, the estimated ̃ diverges further away from the original value. To quantify the difference, the percent change of ̃ and ̃ between the perfect information values (Fig. 4) and the values under uncertainty (Fig. 7) are reported in Table 2 and Table 3. For temperature difference up to = 10℃ , the mean estimation of the damage area does not change significantly with prediction being within 10% of the perfect information case. The temperature difference, however, has a significant impact on the precision of the prediction. Even a = 5℃ difference causes an average increase in the standard deviation by approximately 30% with the situation getting worse at higher temperature difference levels. Table 2: Percent (%) increase in ̃ due to temperature difference. Impact ID = = = = = FP1-1 0.1 2.4 9.1 15.4 31.2 FP2-1 0.5 2.2 10.4 17.2 34.1 FP3-1 0.1 2.6 9.8 16.13 34.3 FP3-2 0.3 2.1 8.3 11.8 22.3 Table 3: Percent (%) increase in ̃ due to temperature difference. Impact ID = = = = = FP1-1 1.8 6.0 20.9 31.4 137 FP2-1 3.3 3.8 11.3 17.2 36.4 FP3-1 1.7 4.9 23.7 41.5 120 FP3-2 3.2 3.6 62.3 161 315 3.2. Influence of Localization Uncertainty The uncertainty in the location estimate can also affect the damage size estimations. As the location of the damage changes, different paths are affected by the existence of damage and therefore the same value corresponds to different values at various locations. In (Giannakeas, Sharif Khodaei, and Aliabadi 2022) the accuracy and precision of the localization has been computed for each impact event. For an impact on the stringer foot, it was estimated that the average accuracy is approximately 17mm while the average precision around 500mm 2 . These results correspond the temperature difference of = 10℃ as reported in (Giannakeas, Sharif Khodaei, and Aliabadi 2022). Therefore, the scenario used for the localization study is conservative as temperature difference is already included. In this section, the influence of location uncertainty on the damage size estimations are studied. As a simplification, = and =0 is used in this study, i.e. it is assumed that the impact coordinate always corresponds at the middle of the stringer foot. This assumption is made because cases where is smaller correspond to delamination damages that do not extend the full width of the stringer foot. Such cases were considered in the numerical simulations, however they were not included in the experimental campaign, and this might introduce considerable bias to the results Table 1. Therefore, only uncertainty on ̃ is included. Furthermore, in this section no temperature difference is assumed and ̃ is set to its true value from Table 1. 663