Ensemble scenario analysis for the statistical performance evaluation of an NDE-based damage prognosis methodology

Developing an integrated structural health monitoring and damage prognosis (SHM-DP) framework for predicting the fatigue life of adhesively-bonded composite joints

“…; qg (Gobbato et al, 2013a;Gobbato, Conte, & Kosmatka, 2013b) and it yields a probabilistic characterization of the future state of damage (or performance level) of the structure at the local reliability component level. ; qg (Gobbato et al, 2013a;Gobbato, Conte, & Kosmatka, 2013b) and it yields a probabilistic characterization of the future state of damage (or performance level) of the structure at the local reliability component level.…”

“…This approach is referred to as "ensemble scenario analysis" and it provides a robust statistical approach for assessing and evaluating the overall performance of the proposed NDE-based SHM-DP methodology (Gobbato et al (2013b). This approach is referred to as "ensemble scenario analysis" and it provides a robust statistical approach for assessing and evaluating the overall performance of the proposed NDE-based SHM-DP methodology (Gobbato et al (2013b).…”

“…The probability of the failure event defined in Eqn (17.22) can be evaluated using state-of-the-art simulation-based techniques such as MCMC methods (Gamerman, 1997;Guan et al, 2011Guan et al, , 2012 and transitional MCMC methods (Ching & Chen, 2007) as well as semi-analytical approaches (Gobbato, 2011;Gobbato et al, 2013aGobbato et al, , 2013b; both types of methods can in fact be embedded into the proposed SHM-DP framework. The probability of the failure event defined in Eqn (17.22) can be evaluated using state-of-the-art simulation-based techniques such as MCMC methods (Gamerman, 1997;Guan et al, 2011Guan et al, , 2012 and transitional MCMC methods (Ching & Chen, 2007) as well as semi-analytical approaches (Gobbato, 2011;Gobbato et al, 2013aGobbato et al, , 2013b; both types of methods can in fact be embedded into the proposed SHM-DP framework.…”

“…However, the recursive Bayesian inference and the debonding evolution analysis represent the two most important steps of the overall methodology to provide accurate and reliable RFL predictions. Examples of componentlevel validation studies can be found in Guan et al (2011Guan et al ( , 2012 and Gobbato et al (2013aGobbato et al ( , 2013b. On the other hand, the effectiveness, accuracy, and robustness of the predictive debonding propagation analysis step rely on the use of calibrated and validated mechanics-based models capable of capturing the physics of the damage propagation processes and the accurate representation of the statistical correlation between the parameters of such models.…”

“…On the one hand, the effectiveness of the recursive Bayesian inference scheme in detecting and quantifying damage is primarily governed by the following factors: detectability and measurement accuracy of the NDE inspection technique used in the monitoring process (see Section 17.3), the number of measurements taken at each inspection, and the inter-inspection time (Gobbato, 2011). Furthermore, it can be proven and verified (Gobbato et al, 2013a(Gobbato et al, , 2013b) that higher levels of uncertainty (e.g., measurement uncertainty, load uncertainty) and fewer inspection data lead toward a flatter pattern of the function G sys t; t p ; d ½0;p m . These two critical aspects often require a considerable amount of experimental investigations (see Section 17.5).…”