BAM Dissertationsreihe
Filtern
Dokumenttyp
- Dissertation (3)
Referierte Publikation
- nein (3)
Schlagworte
- Zuverlässigkeit (3) (entfernen)
Organisationseinheit der BAM
- 7 Bauwerkssicherheit (1)
- 7.2 Ingenieurbau (1)
168
The current practice of operating and maintaining deteriorating structural systems ensures acceptable levels of structural reliability, but it is not clear how efficient it is. Changing the current prescriptive approach to a risk-based approach has great potential to enable a more efficient management of such systems. Risk-based optimization of operation and maintenance strategies identifies the strategy that optimally balances the cost for controlling deterioration in a structural system with the achieved risk reduction. Inspections and monitoring are essential parts of operation and maintenance strategies. They are typically performed to reduce the uncertainty in the structural condition and inform decisions on future operation and maintenance actions. In risk-based optimization of operation and maintenance strategies, Bayesian updating is used to include information contained in inspection and monitoring data in the prediction of the structural reliability. All computations need to be repeated many times for different potential inspection and monitoring outcomes. This motivates the development of robust and efficient approaches to this computationally challenging task.
The reliability of deteriorating structural systems is time-variant because the loads on them and their capacities change with time. In most practical applications, the reliability analysis of deteriorating structural systems can be approached by dividing their lifetime into discrete time intervals. The time-variant reliability problem can then be represented by a series of time-invariant reliability problems. Using this methodology as a starting point, this thesis proposes a novel approach to compute the time-variant reliability of deteriorating structural systems for which inspection and monitoring data are available. The problem is formulated in a nested way in which the prediction of the structural condition is separated from the computation of the structural reliability conditional on the structural condition. Information on the structural condition provided by inspections and monitoring is included in the reliability assessment through Bayesian updating of the system deterioration model employed to predict the structural condition. The updated system reliability is obtained by coupling the updated deterioration model with a probabilistic structural model utilized to calculate the failure probability conditional on the structural condition. This approach is the first main outcome of this thesis and termed nested reliability analysis (NRA) approach. It is demonstrated in two numerical examples considering inspected and monitored steel structures subject to high-cycle fatigue.
An alternative – recently developed – approach, which also follows the strategy of discretizing time, describes deteriorating structural systems with hierarchical dynamic Bayesian networks (DBN). DBN combined with approximate or exact inference algorithms also enable the computation of the time-variant reliability of deteriorating structural systems conditional on information provided by inspection and monitoring data. In this thesis – as a proof of concept – a software prototype is developed based on the DBN approach, which can be used to assess the reliability of a corroding concrete box girder for which half-cell potential measurements are available. This is the second main outcome of this thesis.
Both approaches presented in this thesis enable an integral reliability analysis of inspected and monitored structures that accounts for system effects arising from (a) the correlation among deterioration states of different structural elements, (b) the interaction between element deterioration and system failure, and (c) the indirect information gained on the condition of all unobserved structural elements from inspecting or monitoring the condition of some structural elements. Thus, both approaches enable a systemwide risk-based optimization of operation and maintenance strategies for deteriorating structural systems.
The NRA approach can be implemented relatively easily with subset simulation, which is a sequential Monte Carlo method suitable for estimating rare event probabilities. Subset simulation is robust and considerably more efficient than crude Monte Carlo simulation. It is, however, still sampling-based and its efficiency is thus a function of the number of inspection and monitoring outcomes, as well as the value of the simulated event probabilities. The current implementation of the NRA approach performs separate subset simulation runs to estimate the reliability at different points in time. The efficiency of the NRA approach with subset simulation can be significantly improved by exploiting the fact that failure events in different years are nested. The lifetime reliability of deteriorating structural systems can thus be computed in reverse chronological order in a single subset simulation run.
The implementation of the DBN approach is much more demanding than the implementation of the NRA approach but it has two main advantages. Firstly, the graphical format of the DBN facilitates the presentation of the model and the underlying assumptions to stakeholders who are not experts in reliability analysis. Secondly, it can be combined with exact inference algorithms. In this case, its efficiency neither depends on the number of inspection and monitoring outcomes, nor on the value of the event probabilities to be calculated. However, in contrast to the NRA approach with subset simulation, the DBN approach with exact inference imposes restrictions on the number of random variables and the dependence structure that can be implemented in the model.
153
Die zerstörungsfreie Prüfung (ZfP) ist aus solchen Bereichen unseres Lebens nicht mehr wegzudenken, in denen Schäden mit hohen Folgekosten oder Gefährdungen von Menschenleben entstehen können (Beispiele: Transportwesen, Energieerzeugung, Chemieindustrie). In der Praxis kann ein Prüfsystem an seine Grenzen geraten, z.B. bei kleinen Defekten. Defekte mit kritischer Größe werden möglicherweise nicht detektiert. Daher müssen probabilistische Bewertungsverfahren das Prüfsystem beschreiben.
Es wird eine objektive Qualitätskennzahl gesucht, auf deren Basis die Anwendbarkeit der Prüfmethode definiert werden soll. Die Auffindwahrscheinlichkeit (engl. probability of detection - POD) erfüllt diese Anforderung. Die POD zeigt auf Basis des Zusammenhangs und der Streuung der Daten, ob das Verfahren für die Prüftätigkeit akzeptiert werden kann oder noch verbessert werden muss.
Das ursprüngliche POD-Verfahren wurde für quasi eindimensionale Defekte in dünnen Luftfahrtbauteilen entwickelt. In der industriellen Realität ist diese Bewertung ein Balanceakt zwischen Statistik und Durchführbarkeit: Die Prüfung soll mit realen Defektdaten für die spätere Produktion des Bauteils (bzw. wiederkehrende Wartungsprüfung) bewertet werden. Doch die notwendige Gegenüberstellung zwischen Schliffdaten, für die Erfassung der wahren Defektgröße von räumlich ausgeprägten Defekten und dem Signal eines ZfP-Systems stellt sich als herausfordernde und kostenintensive Aufgabe heraus. Sowohl die Aufstellung eines gemeinsamen Koordiantensystems als auch die Beschreibung und Angleichung der Daten stellen eine notwendige Vorarbeit dar. In dieser Arbeit wird ein mögliches Vorgehen entwickelt, dass im Weiteren eingesetzt werden kann. Während in der Literatur zum Thema POD häufig die Begrenzung des Einsatzes einer eindimensionalen POD (POD mit einem Defektparameter) für reale Defekte bereits erkannt wurde, soll außerdem in dieser Arbeit das Verfahren auf der Signalseite umfassender erweitert werden, um die Einbeziehung realer Defekte in die POD-Bewertung zu ermöglichen. Hierfür werden mit Hilfe dieser Arbeit zwei wesentliche Neuerungen in der POD-Bewertung eingeführt:
1. Die Anzeigenfläche wird als wichtiges Indiz zur Detektion in die Bewertung eingeführt. Dabei zeigt der Ansatz einer Observer-POD, bei dem der Detektierbarkeit eines Defekts beschrieben wird, eine Möglichkeit in die Bewertung zu erweitern. Jedoch wird die notwendige Datenanzahl die für eine Observer-POD selten mit Experimenten erreicht. Daher schlagen wir die Einführung eines Glättungsalgorithmus vor, um auch auf der Basis von wenigen Daten die Flächenabhängigkeit zu erfassen. Der Algorithmus wird hierbei durch simulierte Daten auf seine Funktionsfähigkeit überprüft, bevor er auf reale Defekte angewendet wird. Gleichzeitig helfen die simulierten Daten einen Vergleich zu den vorhergegangenen Ansätzen zu ermöglichen.
2. Darüber hinaus reichen die Daten der realen Defekte häufig nicht aus, um die statistische Forderung zu gewährleisten, so dass es notwendig, wird künstliche Defekte mit einzubeziehen. Deshalb sollen die vorhanden künstlichen Defekte in Form von Referenzdefekten mit einbezogen werden, um die statistische Grundlage zu erhöhen. Für die Prüfung von Referenzdefekten sind jedoch wichtige Einflussgrößen (z.B. Oberflächenrauhigkeit) nicht vorhanden. Wegen der unterschiedlichen Aussagekraft der Daten und zur Vermeidung einer zu optimistischen Abschätzung, ist eine einfache Mischung der Daten ausgeschlossen. Um realen Defekten eine Möglichkeit dafür zu schaffen, dass die Eigenschaften der realen Defekte angemessen auf das Ergebnis der Bewertung des Verfahrens Einfluss nehmen können, wird eine gewichtete Kombination der Defektdaten für die Bewertung vorgestellt. Das Vorgehen wird am Beispiel der radiographischen Prüfung einer elektronenstrahlgeschweißten Naht durchgeführt. Die Schweißnaht verbindet den Deckel zur Außenwand eines Kupferbehältern, der für die spätere Endlagerung von verbrauchten Brennstäben aus Kernkraftwerken entwickelt wurde. Die Messergebnisse stammen aus von der Firma Posiva Oy, dem zuständigen Unternehmen für die Endlagerung von verbrauchten Brennstäben aus Kernkraftwerken in Finnland. Hierbei stellt die POD-Bewertung ein wichtiges Element in der Gesamtrisikobewertung für das Endlagersystem dar.
145
Non-destructive testing (NDT) is regarded as one of the key elements in ensuring quality of engineering systems and their safe use. A failure of NDT to detect critical defects in safetyrelevant components, such as those in the nuclear industry, may lead to catastrophic consequences for the environment and the people. Therefore, ensuring that NDT methods are capable of detecting all critical defects, i.e. that they are reliable, is of utmost importance. Reliability of NDT is affected by human factors, which have thus far received the least amount of attention in the reliability assessments. With increased use of automation, in terms of mechanised testing (automation-assisted inspection and the corresponding evaluation of data), higher reliability standards are believed to have been achieved. However, human inspectors, and thus human factors, still play an important role throughout this process, and the risks involved in this application are unknown. The overall aim of the work presented in this dissertation was to explore for the first time the risks associated with mechanised NDT and find ways of mitigating their effects on the inspection performance. Hence, the objectives were to (1) identify and analyse potential risks in mechanised NDT, (2) devise measures against them, (3) critically address the preventive measures with respect to new potential risks, and (4) suggest ways for the implementation of the preventive measures. To address the first two objectives a risk assessment in form of a Failure Modes and Effects Analysis (FMEA) was conducted (Study 1). This analysis revealed potential for failure during both the acquisition and evaluation of NDT data that could be assigned to human, technology, and organisation. Since the existing preventive measures are insufficient to defend the system from identified failures, new preventive measures were suggested. The conclusion of the study was that those preventive measures need to be carefully considered with respect to new potential risks, before they can be implemented, thus serving as a starting point for further empirical studies. To address the final two objectives, two preventive measures, i.e. human redundancy and the use of automated aids in the evaluation of NDT data, were critically assessed with regard to potential downfalls arising from the social interaction between redundant individuals and the belief in the high reliability of automated aids. The second study was concerned with the potential withdrawal of effort in sequential redundant teams when working collectively as opposed to working alone, when independence between the two redundant individuals is not present. The results revealed that the first redundant inspector, led to believe someone else will conduct the same task afterwards, invested the same amount of effort as when working alone. The redundant checker was not affected by the information about the superior experience of his predecessor and—instead of expected withdrawal of effort—exhibited better performance in the task. Both results were in contradiction to the hypotheses, the explanations for which can be found in the social loafing and social compensation effects and in the methodological limitations. The third study examined inappropriate use of the aid measured in terms of (a) agreement with the errors of the aid in connection to the frequency of verifying its results and in terms of (b) the overall performance in the task. The results showed that the information about the high reliability of the aid did not affect the perception of that aid’s performance and, hence, no differences in the actual use of the aid were to be expected. However, the participants did not use the aid appropriately: They misused it, i.e. agreed with the errors committed by the aid and disused it, i.e. disagreed with the correct information provided by the aid, thereby reducing the overall reliability of the aid in terms of sizing ability. Whereas aid’s misuse could be assigned to low propensity to take risks and reduced verification behaviour because of a bias towards automation, the disuse was assigned to the possible misunderstanding of the task. The results of these studies raised the awareness that methods used to increase reliability and safety, such as automation and human redundancy, can backfire if their implementation is not carefully considered with respect to new potential risks arising from the interaction between individuals and complex systems. In an attempt to minimise this risk, suggestions for their implementation in the NDT practice were provided.