Angewandte Physik
Filtern
Dokumenttyp
- Vortrag (8)
- Zeitschriftenartikel (4)
- Beitrag zu einem Tagungsband (3)
- Posterpräsentation (2)
Sprache
- Englisch (17) (entfernen)
Schlagworte
- Structural Health Monitoring (6)
- Hydrogen (3)
- Machine learning (3)
- Structural health monitoring (3)
- COPV (2)
- Component test (2)
- Composite Overwrapped Pressure Vessel (2)
- Composites (2)
- Damage localization (2)
- Guided Waves (2)
Organisationseinheit der BAM
- 8.4 Akustische und elektromagnetische Verfahren (17) (entfernen)
Eingeladener Vortrag
- nein (8)
Early detection of fatigue cracks and accurate measurements of the crack growth play an important role in the maintenance and repair strategies of steel structures exposed to cyclic loads during their service life. Observation of welded connections is especially of high relevance due to their higher susceptibility to fatigue damage. The aim of this contribution was to monitor fatigue crack growth in thick welded specimens during fatigue tests as holistically as possible, by implementing multiple NDT methods simultaneously in order to record the crack initiation and propagation until the final fracture. In addition to well-known methods such as strain gauges, thermography, and ultrasound, the crack luminescence method developed at the Bundesanstalt für Materialforschung und -prüfung (BAM), which makes cracks on the surface particularly visible, was also used. For data acquisition, a first data fusion concept was developed and applied in order to synchronize the data of the different methods and to evaluate them to a large extent automatically. The resulting database can ultimately also be used to access, view, and analyze the experimental data for various NDT methods. During the conducted fatigue tests, the simultaneous measurements of the same cracking process enabled a comprehensive comparison of the methods, highlighting their individual strengths and limitations. More importantly, they showed how a synergetic combination of different NDT methods can be beneficial for implementation in large-scale fatigue testing but also in monitoring and inspection programs of in-service structures - such as the support structures of offshore wind turbines.
The application of composite overwrapped pressure vessels (COPV) to store hydrogen and other compressed gases, especially when operating at high pressures, imposed the need for an innovative and reliable approach to ensure the safe operation of the system. Continuous structural health monitoring (SHM) based on ultrasonic guided waves (GWs) is a promising approach due to the ability of the wave to propagate for long distances and go around complex structures, moreover the high sensitivity to various failure modes such as delamination, matrix cracking and debonding. In this study, we scrutinize the behavior of (GWs) within the COPV using a network of PZT elements that are used for excitation and sensing. A laser doppler vibrometer (LDV) was also used to scan the surface of the vessel in various directions. L(0,1) and L(0,2) were observed in the captured signal. The L(0,2) appeared to be a dominant mode in the COPV and was capable of propagating along the entire length and maintaining a good signal-to-noise ratio. The L(0,2) mode maintained the same phase velocity when it is captured at various excitation angles (0, 45, and 90 degrees). The reduced effect of the orthotropy of the materials on the propagating waves is an important result as it will reduce the complexity in data processing when performing damage identification.
Offshore wind turbines (OWT) are a key factor of the sustainable energy generation of tomorrow. The continuously increasing installation depths and weight of the OWTs require suitable foundation concepts like monopiles or tripods. Typically, mild steels like the S420ML are used with plate thicknesses up to several hundreds of mm causing high restraints in the weld joints. Due to the large plate thickness, submerged arc welding (SAW) with multiple wires is the state-of-the-art welding procedure. As a result of the very high stiffness of the construction, a certain susceptibility for time-delayed hydrogen-assisted cracking (HAC) may occur. The evaluation of crack susceptibility is very complex due to the component size and stiffness of real offshore structures. For this purpose, a near-component test geometry was developed to transfer the real stiffness conditions to laboratory (i.e., workshop) scale. The investigated mock-up, weighing 350 kg, comprised heavy plates (thickness 50 mm, seam length 1,000 m) joined by a 22-pass submerged arc weld. Additional stiffeners simulated the effect of high restraint or shrinkage restraint of the weld. Extreme scenarios of hydrogen absorption during welding were simulated via the use of welding fluxes in dry (HD < 5 ml/100g Fe) and moisture condition (HD > 15 ml/100g Fe). The residual stresses were determined by a robot X-ray diffractometer. Areas of critical tensile residual stress (at the level of the yield strength) were found in the weld metal and heat affected zone. To identify possible delayed cracking, the welds were tested by phased array ultrasonic testing (PAUT) after 48 h. Summarized, no significant occurrence of HAC was detected, indicating the high crack resistance of the welded joint, i.e., a suitable combination of base material, welding consumable and welding parameters.
Structural Health Monitoring (SHM) mittels geführter Ultraschallwellen ermöglicht eine kontinuierliche Überwachung komplexer Bauteile und liefert umfangreiche Informationen über deren strukturelle Integrität. Die aufgezeichneten Signale sind jedoch komplex und werden daher zunehmend mittels KI verarbeitet, um Anomalien in der Systemintegrität automatisch zu erkennen. Eine Herausforderung dabei ist jedoch das Trainieren solcher KI-Modelle, da dafür große Mengen an Trainingsdaten inklusive diverser Schadensfälle benötigt werden, um eine hohe Genauigkeit und Zuverlässigkeit zu gewährleisten. Experimentelle Generierung dieser Daten ist wegen des enormen Zeit- und Kostenaufwands nicht möglich.
Eine Alternative hierfür bieten die numerischen Simulationen, da dadurch Trainingsdaten für beliebige Anzahl an Schadenskonfigurationen erzeugt werden können. Das Hauptproblem dabei ist jedoch die Abweichung zwischen den experimentellen und simulierten Daten (durch z.B. Abweichungen bei der Geometrie, Randbedingungen oder Materialparameter). Um diese Abweichungen zu reduzieren bzw. zu eliminieren, wird ein KI-Modell mit einer Autoencoder-Decoder Architektur vorgestellt, die mittels eines Experiments kalibriert werden kann und anschließend in der Lage ist, die Genauigkeit der Simulationsergebnisse nachträglich zu erhöhen.
The applications of composite overwrapped pressure vessels (COPVs) in extreme conditions, such as storing hydrogen gases at very high pressure, impose new requirements related to the system's integrity and safety. The development of a structural health monitoring (SHM) system that allows for continuous monitoring of the COPVs provides rich information about the structural integrity of the component. Furthermore, the collected data can be used for different purposes such as increasing the periodic inspection intervals, providing a remaining lifetime prognosis, and also ensuring optimal operating conditions. Ultimately this information can be complementary to the development of the envisioned digital twin of the monitored COPVs. Guided waves (GWs) are preferred to be used in continuous SHM given their ability to travel in complex structures for long distances. However, obtained GW signals are complex and require advanced processing techniques. Machine learning (ML) is increasingly utilized as the main part of the processing pipeline to automatically detect anomalies in the system's integrity. Hence, in this study, we are scrutinizing the potential of using ML to provide continuous monitoring of COPVs based on ultrasonic GW data. Data is collected from a network of sensors consisting of fifteen Piezoelectric (PZT) wafers that were surface mounted on the COPV. Two ML algorithms are used in the automated evaluation procedure (i) a long short-term memory (LSTM) autoencoder for anomaly detection (defects/impact), and (ii) a convolutional neural network (CNN) model for feature extraction and classification of the artificial damage sizes and locations. Additional data augmentation steps are introduced such as modification and addition of random noise to original signals to enhance the model's robustness to uncertainties. Overall, it was shown that the ML algorithms used were able to detect and classify the simulated damage with high accuracy.
While hydrogen is one of the most promising energy carriers, the safety of hydrogen storage technology remains one of the most important factors for technological and societal approval. While the engineering safety factors of the pressure vessels are kept high, the periodic inspection and the limited lifetime are making the application very costly considering manpower, time, money, and material waste. The development of an integrated structural health monitoring system can allow an easy transition from the current situation to cost-effective predictive maintenance. Hence, we propose to integrate three different SHM systems into hydrogen pressure vessels, namely guided wave ultrasonics, acoustic emission, and fibre optic sensing, to continuously monitor the condition and integrity.
In this work, we evaluated the condition of a Type IV composite overwrapped pressure vessel using ultrasonic guided wave propagation. We mounted fifteen piezo-electric wafers on the composite cylinder by shaping three rings containing five sensors each. We acquired data from the sensor network following different boundary conditions with artificial damages on the selected locations. The data were evaluated with guided wave tomography techniques using ultrasonic features (amplitude, frequency, etc.) as well as artificial intelligence (AI). The results suggest that both traditional guided wave fusion techniques and AI-based characterization methods can detect artificial damages. In future work, it is planned to integrate acoustic emission and fibre optic sensing. Moreover, the measurement and the test results will be implemented into a digital twin to derive trends and make predictions on the damage propagation as well as the remaining useful lifetime.
This work has received funding from German Ministry of Economic Affairs and Climate Actions within the QI-Digital initiative (www.qi-digital.de).
Water management in proton-exchange membrane fuel cells (PEMFCs) is critical for efficient operation. However, in-situ measurements of the water distribution in an operating PEMFC are challenging. In this study, we investigate key aspects of ultrasonic guided wave (UGW) propagation through the bipolar plate of a PEMFC using 3D scanning laser Doppler vibrometry (3D LDV). Our results provide insights for the design and optimization of UGW-based sensor systems for water detection in PEMFCs.
The deployment of machine learning (ML) and deep learning (DL) in structural health monitoring (SHM) faces multiple challenges. Foremost among these is the insufficient availability of extensive high-quality data sets essential for robust training. Within SHM, high-quality data is defined by its accuracy, relevance, and fidelity in representing real-world structural scenarios (pristine as well as damaged). Although methods like data augmentation and creating synthetic data can add to datasets, they frequently sacrifice the authenticity and true representation of the data. Sharing real-world data encapsulating true structural and anomalous scenarios offers promise. However, entities are often reluctant to share raw data, given the potential extraction of sensitive information, leading to trust issues among collaborating entities.
Our study introduces a novel methodology leveraging Federated Learning (FL) to navigate these challenges. Within the FL framework, models are trained in a decentralized manner across different entities, preserving data privacy. In our research, we simulated several scenarios and compared them to traditional local training methods. Employing guided wave (GW) datasets, we distributed the data among different parties (clients) using IID (independent, identically distributed or in other words, statistically identical) mini batches of dataset, as well as non-IID configurations. This approach mirrors real-world data distribution among varied entities, such as hydrogen refueling stations.
In our methodology, the initial round involves individualized training for each client using their unique datasets . Subsequently, the model parameters are sent to the FL server, where they are averaged to construct a global model. In the second round, this global model is disseminated back to the clients to aid in predictive tasks. This iterative process continues for several rounds to reach convergence.
Our findings distinctly highlight the advantages of FL over localized training, evidenced by a marked improvement in prediction accuracy . This research underscores the potential of FL in GW-based SHM, offering a remedy to similar challenges tied to data scarcity in other SHM approaches and paving the way for a new era of collaborative, data-centric monitoring systems.
Offshore wind turbines continuously increase in size and weight and demand adequate offshore foundations concepts like monopiles, tripods, or jackets. These components are typically constructed using submerged arc welding (SAW) with high-strength thick steel plates like the S420ML. During welding, the occurrence of delayed hydrogen-assisted cracking (HAC) must be anticipated. HAC is a critical combination of the local hydrogen concentration within a susceptible microstructure under certain mechanical load, i.e., the occurring (welding) residual stresses. The welding sequence of the thick-walled plates complicates the residual stress distribution due to the necessary repeated thermal cycling, i.e., welding seam/layer deposition to fill the joint. For that purpose, SAW with two-wire-technique was used to weld a specially designed and prototype-like mock-up of a real component with a thickness of 50 mm, filled with over 20 passes and a seam length of 1000 mm. Additional welded stiffeners simulated the effect of a high restraint, to achieve critical HAC conditions. The necessity of a minimum waiting time (MWT) before the NDT can be conducted (to exclude HAC) was critically verified by the application of ultrasonic testing of the welded joint at different time-steps of the NDT of up to 48 h after the completion welding. The residual stresses were determined by a robot XRD goniometer. Tensile residual stresses up to the yield limit are found both in the weld metal and in the heat-affected zone. Numerical modeling allowed the qualitative estimation of the hydrogen diffusion in the weld. No noticeable HAC occurrence was identified and confirms the high cracking resistance of the investigated material. Finally, the applicability of the MWT concept should be critically discussed.
The application of composite overwrapped pressure vessels (COPV) to store hydrogen and other compressed gases, especially when operating at high pressures, imposed the need for an innovative and reliable approach to ensure the safe operation of the system. Continuous structural health monitoring (SHM) based on ultrasonic guided waves (GWs) is a promising approach due to the ability of the wave to propagate for long distances and go around complex structures, moreover the high sensitivity to various failure modes such as delamination, matrix cracking and debonding. In this study, we scrutinize the behavior of (GWs) within the COPV using a network of PZT elements that are used for excitation and sensing. A laser doppler vibrometer (LDV) was also used to scan the surface of the vessel in various directions. L(0,1) and L(0,2) were observed in the captured signal. The L(0, 2) appeared to be a dominant mode in the COPV and was capable of propagating along the entire length and maintaining a good signal-to-noise ratio. The L(0,2) mode maintained the same phase velocity when it is captured at various excitation angles (0, 45, and 90 degrees). The reduced effect of the orthotropy of the materials on the propagating waves is an important result as it will reduce the complexity in data processing when performing damage identification.