Filtern
Dokumenttyp
- Zeitschriftenartikel (3)
- Buchkapitel (2)
- Vortrag (1)
- Posterpräsentation (1)
Sprache
- Englisch (7)
Schlagworte
- X-ray computed tomography (4)
- Deep learning (3)
- Deep Learning (2)
- Very high-cycle fatigue (2)
- Additive manufacturing (1)
- Aluminum (1)
- Compression (1)
- Data Compression (1)
- Defects (1)
- Digital Twin (1)
Organisationseinheit der BAM
We present a 3D image registration technique for non-linear deformation estimation in Additive Manufacturing processes. The methodology involves comparing X-ray Computed Tomography (XCT) data with Computer Aided Design (CAD) models for Triply Periodic Minimal Surface (TPMS) lattices and employs the Chamfer distance to refine mesh non-linear deformations.
X‐ray computed tomography (XCT) is extremely useful for the non‐destructive analysis of additively manufactured (AM) components. AM components often show manufacturing defects such as lack‐of‐fusion (LoF), which are detrimental to the fatigue life of components. To better understand how cracks initiate and propagate from internal defects, we fabricated Ti‐6Al‐4V samples with an internal cavity using electron beam powder bed fusion. The samples were tested in high‐cycle and very high‐cycle fatigue regimes. XCT was used to locate crack initiation sites and to determine characteristic properties of cracks and defects with the aid of deep learning segmentation. LoF defects exposed to the outer surface of the samples after machining were found to be as detrimental to fatigue life as the internal artificial defects. This work can benefit industries that utilize the AM of high‐strength, lightweight alloys, in the design and manufacturing of components to improve part reliability and fatigue life.
X-ray computed tomography (XCT) is an invaluable method for evaluating the properties and performance of components both during service and after failure in a non-destructive manner. XCT is particularly useful for the investigation of additively manufactured (AM) components, which often have production defects that are inherent to the manufacturing process, such as lack of fusion defects. Understanding the mechanisms of fatigue crack growth throughout the life cycle of such components is crucial and so to address this need, we designed and performed experiments to investigate the fatigue life and fatigue crack growth behavior of Ti-6Al-4V components under very high cycle fatigue (VHCF) testing. The titanium samples were additively manufactured with intentional internal defects to control crack initiation location. XCT of the component was carried out to identify crack initiation sites and characterize the dynamics of crack growth. The findings from this work will benefit industries that rely on the AM of titanium alloys, aiding in the improvement of component design and manufacturing processes.
The ever-growing volume of data in imaging sciences stemming from advancements in imaging technology, necessitates efficient and reliable storage solutions for such large datasets. In this study, we investigated the compression of industrial X-ray computed tomography (XCT) data using deep learning and examined how these compression algorithms affect the quality of the recovered data. To determine best strategies, we investigated two different network architectures with different compression rates. The XCT data used was from a sandstone sample with a complex internal pore network. The quality of the recovered images obtained from the two different deep learning architectures with different compression rates were quantified and compared to the original input data. We showed that the most suitable architecture and compression rate can change, depending on what specific characteristics we would need to preserve for later analysis. The findings presented here can aid scientists in determining the best approaches and strategies for their data storage and analysis requirements.
Since the introduction of high-strength aluminum alloys understanding their fatigue behavior is of high interest for the structural integrity of engineering components, in this context, the alloy EN AW-2618A gains its high strength from both nanometer size precipitates and micrometer size primary phases. The latter phases are often identified as crack initiation sites. In this study, it uses a combination of synchrotron and laboratory-based X-ray computed tomography to image and quantify such primary phases and the fatigue cracks appearing in interrupted tests. Based on the gray-level differences in the synchrotron X-ray computed tomography scans, this study is able to distinguish low- and high-absorbing particles. The dominant (volume fraction >99%) high absorbing primary phase can be quantified in good agreement with results of Thermo-Calc calculations. To image the fatigue crack, laboratory X-ray computed tomography scans are performed at different tensile loads to open the crack. The results show that with an appropriate crack opening tensile load, the fatigue crack morphology can be reliably revealed. Based on these results, the influence of the primary phases on the fatigue crack initiation and propagation are discussed.
Quality control in Additive Manufacturing (AM) is vital for industrial applications in areas such as the automotive, medical, and aerospace sectors. Geometric inaccuracies caused by shrinkage and deformations can compromise the life and performance of additively manufactured components. Such deviations can be quantified using Digital Volume Correlation (DVC), which compares the Computer-Aided Design (CAD) model with the X-ray Computed Tomography (XCT) geometry of the components produced. However, accurate registration between the two modalities is challenging due to the absence of a ground truth or reference deformation field. In addition, the extremely large data size of high-resolution XCT volumes makes computation difficult. In this work, we present a deep learning-based approach for estimating voxel-wise deformations between CAD and XCT volumes. Our method uses a dynamic patch-based processing strategy to handle high-resolution volumes. In addition to the Dice score, we introduce a Binary Difference Map (BDM) that quantifies voxel-wise mismatches between binarized CAD and XCT volumes to evaluate the accuracy of the registration. Our approach shows a 9.2% improvement in the Dice score and a 9.9% improvement in the voxel match rate compared to classic DVC methods, while reducing the interaction time from days to minutes. This work sets the foundation for deep learning-based DVC methods to generate compensation meshes that can then be used in closed-loop correlations during the AM production process. Such a system would be of great interest to industry, as it would make the manufacturing process more reliable and efficient, saving time and material.
The ever-growing volume of data in imaging sciences stemming from advancements in imaging technologies, necessitates efficient and reliable storage solutions for such large datasets. This study investigates the compression of industrial X-ray computed tomography (XCT) data using deep learning autoencoders and examines how these compression algorithms affect the quality of the recovered data. Two network architectures with different compression rates were used, a deep convolution neural network (D-CNN) and a vector quantized variational autoencoder (VQ-VAE). The XCT data used was from a sandstone sample with a complex internal pore network as a good test case for the importance of feature preservation. The quality of the decoded images obtained from the two different deep learning architectures with different compression rates were quantified and compared to the original input data. In addition, to improve image decoding quality metrics, we introduced a metric sensitive to edge preservation, which is crucial for three-dimensional data analysis. We showed that different architectures and compression rates are required depending on the specific characteristics needed to be preserved for later analysis. The findings presented here can aid scientists in determining the requirements and strategies needed for appropriate data storage and analysis.
