Considering components subjected to fatigue loads, in presence of Stress Intensity Factors higher than design assumptions, it is licit to expect crack nucleation and consequent propagation during service. Crack nucleation sites, in the most critical design sections, are expected where manufacturing, or service (also related to aggressive environments) defects are present. From this point of view, it is possible to employ the “Damage Tolerant” design approach whose philosophy consists ([1],[2]) in determining the most opportune inspection interval given the “Probability of Detection” (POD) curve ([3],[4]) of the adopted Non-Destructive Testing (NDT) technique or, alternatively, in defining the needed NDT specifications given a programmed inspection interval. Structural integrity of safety components during service is then strictly related to different factors such as: i) the performance of non destructive testing (“NDT”); ii) the crack propagation behaviour of the adopted material; iii) the reliable knowledge of service loads. Up till now, the performance of a given NDT technique in detecting cracks and defects has been quantified and summarised by the corresponding POD curve relating the probability to detect the defect to its characteristic linear dimension (diameter, length, depth, etc.). Recent results published by the authors ([5]-[7]) showed that such “linear dimension” approach is not the most effective in order to categorise the response of artificial and natural defects having different dimensions and shapes. In particular, artificial defects obtained by traditional machining (saw-cuts) and EDM (concave and convex shaped) on sample blocks obtained from railway axles and fatigue cracks propagated into full-scale axles were considered and compared. Better results in correlating ultrasonic echoes, for all the shapes, dimensions and artificial or natural defects, could be achieved adopting, as a characteristic parameter, the reflecting area. The present paper, based onto such results added with more experimental evidences, proposes a novel and more efficient approach to the definition of POD curves in terms of reflecting area of defects instead of their linear dimension. The application and the usefulness of the novel approach will be here exemplified considering the relevant case of hollow railway axles.
A NEW APPROACH FOR THE DEFINITION OF “PROBABILITY OF DETECTION” CURVES
CARBONI, MICHELE;
2010-01-01
Abstract
Considering components subjected to fatigue loads, in presence of Stress Intensity Factors higher than design assumptions, it is licit to expect crack nucleation and consequent propagation during service. Crack nucleation sites, in the most critical design sections, are expected where manufacturing, or service (also related to aggressive environments) defects are present. From this point of view, it is possible to employ the “Damage Tolerant” design approach whose philosophy consists ([1],[2]) in determining the most opportune inspection interval given the “Probability of Detection” (POD) curve ([3],[4]) of the adopted Non-Destructive Testing (NDT) technique or, alternatively, in defining the needed NDT specifications given a programmed inspection interval. Structural integrity of safety components during service is then strictly related to different factors such as: i) the performance of non destructive testing (“NDT”); ii) the crack propagation behaviour of the adopted material; iii) the reliable knowledge of service loads. Up till now, the performance of a given NDT technique in detecting cracks and defects has been quantified and summarised by the corresponding POD curve relating the probability to detect the defect to its characteristic linear dimension (diameter, length, depth, etc.). Recent results published by the authors ([5]-[7]) showed that such “linear dimension” approach is not the most effective in order to categorise the response of artificial and natural defects having different dimensions and shapes. In particular, artificial defects obtained by traditional machining (saw-cuts) and EDM (concave and convex shaped) on sample blocks obtained from railway axles and fatigue cracks propagated into full-scale axles were considered and compared. Better results in correlating ultrasonic echoes, for all the shapes, dimensions and artificial or natural defects, could be achieved adopting, as a characteristic parameter, the reflecting area. The present paper, based onto such results added with more experimental evidences, proposes a novel and more efficient approach to the definition of POD curves in terms of reflecting area of defects instead of their linear dimension. The application and the usefulness of the novel approach will be here exemplified considering the relevant case of hollow railway axles.File | Dimensione | Formato | |
---|---|---|---|
Paper.pdf
Accesso riservato
:
Post-Print (DRAFT o Author’s Accepted Manuscript-AAM)
Dimensione
796.56 kB
Formato
Adobe PDF
|
796.56 kB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.