Advanced ultrasonic “Probability of Detection” curves for designing in-service inspection intervals

•A set of seventeen natural fatigue cracks in full-scale hollow axles was built.•Master POD curves, function crack reflecting area, are independent from its shape.•All initial cracks tend to the same shape increasing the number of fatigue cycles.•Cracks with the same reflecting area showed very similar probabilities of failure.•Master POD curves provide lower probabilities of failure with respect to traditional POD ones.

In order to ensure the safe service of modern railway vehicles, safety critical components are subjected to a dedicated maintenance and inspection plan. Railways axles, in particular, are periodically inspected by the ultrasonic testing method during maintenance interruptions in first level workshops, while the magnetic particles test is carried out in second level workshops. The reliability of such inspections is quantified in terms of “Probability of Detection” curves, which are traditionally related to a specific linear dimension of the defect to be detected. Actually, a fatigue crack is known to change its shape under cyclic loading, so affecting its detectability.In the present paper, a novel approach to “Probability of Detection” takes into account of this effect, introducing a “Master” probability of detection curve, which is a function of the reflecting area of the crack. A comparative applicative example between standard and such advanced probabilities of detection curves is then presented considering high-speed railway applications, whose hollow axles are made of the medium strength EA4T steel grade.