Datei im Netzwerk der BAM verfügbar ("Closed Access")
Filtern
Dokumenttyp
- Vortrag (7)
- Zeitschriftenartikel (2)
- Forschungsbericht (1)
Sprache
- Englisch (10)
Schlagworte
- POD (8)
- Reliability (5)
- NDT (3)
- Ultrasonic (3)
- Human factors (2)
- Probability of detection (2)
- Railway (2)
- Bayesian approach (1)
- Copper (1)
- Crack (1)
Eingeladener Vortrag
- nein (7)
Non-Destructive Testing (NDT) is routinely used in aerospace, nuclear, railway and automotive industries. The most common use of NDT is to find flaws in components. In dynamically loaded components, flaws grow, and if they are not detected in time and no corrective measures are taken they can reach critical sizes. When a flaw reaches a critical size, it endangers the structural integrity of the component and will likely cause the component to fail. To detect flaws before they reach a critical size, components are inspected in fixed time intervals with non-destructive testing systems. In order to determine the amount of time between inspections, three variables are necessary: the critical size of the flaw, the flaw propagation speed and the size of the flaw that can be reliably detected by NDT. All three variables are, to a certain extent, stochastic. The critical size of the flaw and flaw propagation speed are questions of fracture mechanics. The size of the flaw that can be reliably detected by NDT is predicted by POD curves. The POD is determined for a given inspection scenario. The inspection scenario includes the NDT system, the component geometry and the type of the flaw. If one of these changes, the POD will also change and needs to be recalculated. As a result, the design of a dynamically loaded safety-relevant components is an iterative process in which a geometry is sought that provides an optimum between structural integrity and inspectability. A model that describes the iterative process for the design of the safe components with examples from praxis will be presented.
Nondestructive testing (NDT) is routinely used in the nuclear, rail, aerospace and automotive industries to search for flaws in components. A signal from the flaw, recorded by the NDT device, will vary from measurement to measurement. The sources of this innate signal variation can be categorized into intrinsic, human and application factors. This variation, especially when searching for flaws that are at the limits of the NDT detection capabilities, can result in a failure to detect a flaw. If the inspected components are safety critical, the capability of NDT system to find flaws must be determined in order to avoid the catastrophic consequences of a missed flaw. The NDT system capability to detect flaws is expressed in terms of reliability. The probability of detection (POD) curve is a widespread tool to quantify the reliability of NDT. The POD is determined by series of experiments on specimens containing a range of flaws with known characteristics. The production of a sufficient number of these flaws is time consuming and expensive. In this paper, a multi-parameter POD model that uses both simulation and experimental measurements to calculate the POD curves will be presented. Simulation is used to assess the intrinsic capability of the NDT system and the variability in the system is estimated from experimental measurements. The POD calculated with the multi-parameter model is more comprehensive than the one calculated with the traditional model and the number of costly experiments needed is reduced.
Today, it is an established fact that the capability of the non-destructive testing (NDT) to find flaws can be properly addressed only in terms of probability of detection (POD). The probabilistic, signal-response model, introduced in 1980s, was developed with experimental observation of eddy-current inspections of flat plate samples, containing surface breaking cracks. A linearity between the peak voltage measured by the testing system, and the crack depth was observed. The influence of the crack depth was therefore seen as the major influencing factor for the POD, whereas other factors merely caused the variability in the measurement. This model has proven itself valid for those inspection cases where there is only one main influencing factor on the POD (usually the flaw size) and other factors have a lesser influence. But with increasing requirements to quantify the capability of NDT systems in complex inspection situations, where several factors have a major influence on the POD, it has become clear that the applicability of this simple model has reached its limits. In disregard to its limitations, this model is regularly applied to those situations in which its fundamental assumptions are invalid, forcing evaluators and NDT researchers into attempts to fit the data to an unsuitable model, instead of fitting an appropriate model to the data.
The multiparameter POD model, developed in the early 2010s, enables more factors that influence detection to be simultaneously analysed, making the POD a function of multiple factors. The model is based on the inspection’s physical model, to describe the influence of different factors on the response signal. Measurement variability is obtained from the experiment. Using this model assisted determination of the POD, the necessary number of flaws for evaluation is reduced and a more comprehensive understanding of the inspection is obtained. Several examples of the successful application of the multi-parameter POD model in different fields will be presented.
Reliability analysis of the ultrasonic inspection system for the inspection of hollow railway axles
(2016)
Axles are safety critical train components that are subjected to significant cyclic loading during operation. If the crack is initiated in the axle, cyclic loading will lead to crack propagation. To maintain structural integrity, axles must be periodically inspected for fatigue cracks in the material. Deutsche Bahn uses mechanized ultrasonic inspection system to inspect hollow railway axles. The inspections are performed from the bore surface, using several conventional transducers with different incident angles, inspecting the axle along the entire length. As with the every safety critical system, the reliability of these inspections must be determined with regard to their flaw detection capabilities. Traditionally this is done according to the relevant standards for railway vehicles. To investigate the capability of the NDT system more thoroughly, we want to evaluate the capability of the inspection system to detect flaws by means of probability of detection (POD) curves. It will be shown that other parameters, beside the size of the crack, for example crack position in the axle, influence the detection of the crack. The influence of these parameters was evaluated using ultrasonic simulation. The evaluation served as an input for the manufacturing of the flaws in the real scale axle. Once these axles are inspected and the data evaluated, using data from both measurement and simulation, we will express the POD of the crack as function of influencing parameters using the multiparameter POD model.
The canister for the permanent storage of spent nuclear fuel used by SKB in Sweden consists of a cast iron insert surrounded by a five centimetre thick shell of copper. It is a safety critical component and in order to secure long-term structural integrity non-destructive methods are used to inspect 100% of the volume of each canister, before it is disposed of in the repository. One of the critical components that requires inspection is a sealing weld, joining the copper tube and the lid. The friction stir weld is inspected using an ultrasonic phased array system. The area of the weld is inspected with several inspection channels with different angles and varying coverage. To make sure that no defects that might occur in the weld are overseen, the reliability of the inspection must be quantified. The reliability of NDT is usually quantified with the probability of detection curves. The influence of the parameters that might influence the POD of the flaws in the weld is investigated analysing the experimental results, as well as with a help of a numerical simulation of the inspection.
Deutsche Bahn periodically inspects hollow railway axles for fatigue cracks on the outer surface with mechanized ultrasonic inspection systems. According to the current standard for the inspection of railway axles, the capability of the inspection system to detect these cracks has to be demonstrated on the saw-cut type artificial defects. However, the geometry and the ultrasonic response of the real cracks that can occur on the outer surface of the axle are different from the saw-cut. Furthermore, it is demonstrated that the position and the orientation of the cracks are also important factors that influence the crack detectability. It is proposed to evaluate the influence of all factors on the detection of the cracks using the multi-parameter reliability model. The model uses numerical simulation and experiments to comprehensively address the influence of several factors on the probability of detection.
Since the beginnings of the European American Workshops (EAW) in 1997, the aim was to gather the experts in NDE reliability and discuss burning topics with the aim of identifying crucial problems and suggesting ways to move forward. This was usually achieved during the so-called break-out sessions, in which predetermined topics were discussed. During the 5th EAW, held in Berlin in 2013, this approach was replaced by an Open Space Technology (OST) approach. The benefit of this approach is seen in the freedom of topic choices, i. e., the topics are not predetermined but rather arise at that moment in that space and by the participants choice. The following topics arose: new reliability methods (Bayesian, MAPOD,
), structural health monitoring, definition of requirements of NDE by customer versus provider, what value of POD is good enough?, human factors, manual versus automated inspection, and basic concepts of reliability of NDE. The participants were encouraged to walk from one session to another and openly express their opinions. These were in the end summarized by a chosen group of moderators and presented in this paper.
This report describes the progress in understanding and describing the detectability of the ultrasonic inspection technique developed by SKB for the inspection of copper tubes used for the final disposal of the Swedish spent nuclear fuel. In former research activities dedicated to the different parts of the canister, the probability of detection (POD) evaluation technique, as developed for thin aircraft components, was further developed for the application on complex hick-walled components. The result of this development was the introduction of a “multi-parameter” POD framework. In contrast to taking only the defect size into consideration (as for aircraft components) additional influencing factors, relevant for thick components, such as depth position, orientation and part geometry, were included. Furthermore, the variation of ultrasonic attenuation due to various material properties was included.
Das Ziel des Artikels ist es eine Übersicht über neue methodische Ansätze und Verfahren zur Bewertung der Zuverlässigkeit von ZfP-Systemen in Übereinstimmung mit den spezifischen Anforderungen verschiedener industrieller Anwendungen zu geben. Nach einem Überblick über die in den vergangenen Jahrzehnten erreichten Fortschritte, wird der Fokus auf die gegenwärtigen Entwicklungen gerichtet. Für hohe Sicherheitsanforderungen wurde die quantitative Fehlerauffindwahrscheinlichkeit (probability of detection, POD) basierend auf 'hit/miss'- oder 'Signal-Response'-Analysen und die ROC-Analyse (Receiver Operating Characteristics) als typische Werkzeuge entwickelt. Im Rahmen des modularen Modells für die ZfP-Zuverlässigkeit unterscheidet man zwischen den rein physikalisch-technischen Einflussfaktoren, den industriellen Anwendungsfaktoren und den menschlichen Faktoren. Es hilft zum Beispiel zu erkennen, welche dieser Faktoren man in Modellierungsrechnungen bzw. mittels offenen oder Blindversuchen untersuchen kann. Ein neues Paradigma wird für die Betrachtung der POD als eine Funktion der gewählten Systemparameter und somit als Optimierungswerkzeug verwendet, anstelle sie für die endgültige Beurteilung zu nehmen. Zur Handhabung der Zuverlässigkeitsuntersuchung von realen Defekten in einer realistischen Umgebung werden ökonomisch vertretbare, aber dennoch präzise genug arbeitende Methoden, wie der Ansatz von Bayes oder modellbasierte Methoden vorgestellt. Die menschlichen Faktoren sind oft von ausschlaggebender Wichtigkeit für die Gesamtzuverlässigkeit. Hier hilft ein systematischer psychologischer Ansatz die Engpässe und Verbesserungsmöglichkeiten zu finden.