Wissenschaftliche Artikel der BAM
Filtern
Erscheinungsjahr
Dokumenttyp
Schlagworte
- Additive manufacturing (23)
- Computed tomography (12)
- Residual stress (10)
- Laser powder bed fusion (7)
- Additive Manufacturing (6)
- Neutron diffraction (6)
- Synchrotron X-ray diffraction (6)
- Ti-6Al-4V (6)
- X-ray refraction (6)
- X-ray computed tomography (5)
Organisationseinheit der BAM
- 8 Zerstörungsfreie Prüfung (74)
- 8.5 Röntgenbildgebung (74)
- 9 Komponentensicherheit (16)
- 9.4 Integrität von Schweißverbindungen (12)
- 5 Werkstofftechnik (11)
- 5.1 Mikrostruktur Design und Degradation (6)
- 9.6 Additive Fertigung metallischer Komponenten (6)
- 8.3 Thermografische Verfahren (5)
- 5.3 Polymere Verbundwerkstoffe (3)
- 5.2 Metallische Hochtemperaturwerkstoffe (2)
- 2 Prozess- und Anlagensicherheit (1)
- 2.5 Konformitätsbewertung Explosivstoffe/Pyrotechnik (1)
- 3 Gefahrgutumschließungen; Energiespeicher (1)
- 3.1 Sicherheit von Gefahrgutverpackungen und Batterien (1)
- 3.6 Elektrochemische Energiematerialien (1)
- 5.0 Abteilungsleitung und andere (1)
- 5.4 Multimateriale Fertigungsprozesse (1)
- 7 Bauwerkssicherheit (1)
- 7.1 Baustoffe (1)
- 8.4 Akustische und elektromagnetische Verfahren (1)
- 9.3 Schweißtechnische Fertigungsverfahren (1)
Quantitative microstructural analysis of XCT 3D images is key for quality assurance of materials and components. In this paper we implement a Graph Convolutional Neural Network (GCNN) architecture to segment a complex Al-Si Metal Matrix composite XCT volume (3D image). We train the model on a synthetic dataset and we assess its performance on both synthetic and experimental, manually-labeled, datasets. Our simple GCNN shows a comparable performance, measured via the Dice score, to more standard machine learning methods, but uses a greatly reduced number of parameters (less than 1/10 of parameters), features low training time, and needs little hardware resources. Our GCNN thus achieves a cost-effective reliable segmentation.
Determination of short carbon fiber orientation in zirconium diboride ceramic matrix composites
(2024)
In fiber-reinforced components, the fiber alignment and orientation have paramount influence on the thermomechanical properties of the resulting composite, for both short and continuous fiber. Here we present the case of an ultra-refractory matrix intended for extreme environment applications, ZrB2, reinforced with 20 vol% and 50 vol% short carbon fibers. In both cases, fibers tend to align perpendicular to the uniaxial pressure applied during shaping and sintering of a pellet, although the fiber tilt across the pellet thickness is difficult to determine. Moreover, for high volume fractions of reinforcement, the spatial distribution of the fibers is heterogeneous and tends to have domains of preferential orientations. We compare the information on the fiber distribution as collected by scanning electron microscopy images, X-ray computed tomography and synchrotron X-ray refraction radiography (SXRR). The three techniques prove to be complementary. Importantly, we demonstrate that SXRR yields the most statistically significant information due to the largest field of view, yet with a sensitivity down to the nanometer, and that can be successfully applied also to heavy matrix materials, such as zirconium boride.
AbstractIn this study, we compare the residual stress state in a laser powder bed fusion (PBF‐LB/M) AlSi10Mg alloy in the as‐built (AB) condition with that after two different heat treatments (265 °C for 1 h, HT1; and 300 °C for 2 h, HT2). The bulk residual stress (RS) is determined using synchrotron X‐ray diffraction (SXRD), and near‐surface profiles are determined using laboratory energy‐dispersive X‐ray diffraction (EDXRD). The EDXRD results do not reveal any notable difference between the conditions at a depth of 350 μm, suggesting that the machining process yields a comparable residual stress state in the near‐surface regions. On the other hand, the SXRD results show that HT1 is more effective in relieving the bulk RS. It is observed that HT1 reduces the RS state in both the aluminium matrix and the silicon network. In addtion, HT2 does not have a significant impact on relaxing the RS as‐built state of the matrix, although it does induce a reduction in the RS magnitudes of the Si phase. It is concluded that the heat treatment stress relieving is effective as long as the Si‐network is not disaggregated.
Powder Bed Fusion – Laser Beam (PBF-LB) of high γ’ strengthened Ni-base superalloys, such as CM247LC, is of great interest for high temperature applications in gas turbines. However, PBF-LB of CM247LC is challenging due to the high cracking susceptibility during PBF-LB processing (solidification cracking) and heat treatment (strain age cracking, mostly caused by residual stresses). This study focuses on understanding the impact of process parameters on microstructure, residual stresses and solidification cracking. Laser power (P), speed (v) and hatch spacing (h) were varied while the layer thickness (t) was fixed. The melt pool size and shape were found to be key factors in minimizing solidification cracking. Narrower and shallower melt pools, achieved using a low line energy density (LED = P/v ≤ 0.1 J/mm), gave low crack densities (0.7 mm/mm2). A tight hatch spacing (h = 0.03 mm) resulted in reduced lack of fusion porosity. Electron backscatter diffraction investigations revealed that parameters giving finer microstructure with 〈100〉crystallographic texture had low crack densities provided they were processed with a low LED. Atom probe tomography elucidated early stages of spinodal decomposition in the as-built condition, where Cr and Al cluster separately. The extent of spinodal decomposition was found to be affected by the LED and the hatch spacing. Samples with low LED and small hatch spacing showed higher degrees of spinodal decomposition. X-ray diffraction residual stress investigations revealed that the residual stress is proportional to the volumetric energy density (VED = P/(v. h. t)). Although low residual stresses can be achieved by using low VED, there is a high risk of lack of fusion. Hence, other parameters such as modified scan strategy, build plate pre-heating and pulsed laser mode, must be further explored to minimize the residual stresses to reduce the strain age cracking susceptibility.
Efficient density evolution during sintering of the as-printed component is vital to reach full densification and required properties of binder jet (BJT) components. However, due to the high porosity and brittle nature of the green compact, analysis of the microstructure development during sintering is very difficult, resulting in lack of understanding of the densification process. Density development from green state (57 ± 1.6 %) up to full density (99 ± 0.3 %) was characterized by high-resolution synchrotron X-Ray computed tomography (SXCT) on BJT 316L samples from ex-situ interrupted sintering tests. Periodicity of density fluctuations along the building direction was revealed for the first time and was related to the layer thickness of ~ 42 μm during printing that decreased down to ~ 33 μm during sintering. Sintering simulations, utilizing a continuum sintering model developed for BJT, allowed to replicate the density evolution during sintering with a mean error of 2 % and its fluctuation evolution from green (1.66 %) to sintered (0.56 %) state. Additionally, simulation of extreme particle size segregation (1 μm to 130 μm) suggested that non-optimized printing could lead to undesirable density fluctuation amplitude rapid increase (~10 %) during sintering. This might trigger the nucleation of defects (e.g., layer delamination, cracking, or excessive residual porosity) during the sintering process.
Simulation-Assisted Augmentation of Missing Wedge and Region-of-Interest Computed Tomography Data
(2024)
This study reports a strategy to use sophisticated, realistic X-ray Computed Tomography (CT) simulations to reduce Missing Wedge (MW) and Region-of-Interest (RoI) artifacts in FBP (Filtered Back-Projection) reconstructions. A 3D model of the object is used to simulate the projections that include the missing information inside the MW and outside the RoI. Such information augments the experimental projections, thereby drastically improving the reconstruction results. An X-ray CT dataset of a selected object is modified to mimic various degrees of RoI and MW problems. The results are evaluated in comparison to a standard FBP reconstruction of the complete dataset. In all cases, the reconstruction quality is significantly improved. Small inclusions present in the scanned object are better localized and quantified. The proposed method has the potential to improve the results of any CT reconstruction algorithm.
AbstractLithium‐ion cells connected in series are prone to an electrical safety risk called overdischarge. This paper presents a comprehensive investigation of the overdischarge phenomenon in lithium‐ion cells using operando nondestructive imaging. The study focuses on understanding the behavior of copper dissolution and deposition during overdischarge, which can lead to irreversible capacity loss and internal short‐circuits. By utilizing synchrotron X‐ray computed tomography (SXCT), the concentration of dissolved and deposited copper per surface area is quantified as a function of depth of discharge, confirming previous findings. The results also highlight for the first time a nonuniform distribution pattern for copper deposition on the cathode. This research provides insights for safer battery cell design.
It is shown that preconditioning of experimental X‐ray computed tomography (XCT) data is critical to achieve high‐precision segmentation scores. The challenging experimental XCT datasets and deep convolutional neural networks (DCNNs) are used that are trained with low‐resemblance synthetic XCT data. The material used is a 6‐phase Al–Si metal matrix composite‐reinforced with ceramic fibers and particles. To achieve generalization, in our past studies, specific data augmentation techniques were proposed for the synthetic XCT training data. In addition, two toolsets are devised: (1) special 3D DCNN architecture (3D Triple_UNet), slicing the experimental XCT data from multiple views (MultiView Forwarding), the i.S.Sy.Da.T.A. iterative segmentation algorithm, and (2) nonlocal means (NLM) conditioning (filtering) for the experimental XCT data. This results in good segmentation Dice scores across all phases compared to more standard approaches (i.e., standard UNet architecture, single view slicing, standard single training, and NLM conditioning). Herein, the NLM filter is replaced with the deep conditioning framework BAM SynthCOND introduced in a previous publication, which can be trained with synthetic XCT data. This leads to a significant segmentation precision increase for all phases. The proposed methods are potentially applicable to other materials and imaging techniques.
In the present study, in-situ observation of Hot Isostatic Pressure (HIP) procedure of laser powder bed fusion manufactured Ti-6Al-4V parts was performed to quantitatively estimate the densifcation rate of the material and the infuence of the defect initial size and shape on such rate. The observations were performed in-situ using the Ultrafast Tomography Paris-Edinburgh Cell and the combination of fast phase-contrast synchrotron X-ray tomography and energy dispersive difraction. With this strategy, we could quantify how the efectiveness of HIP depends on the characteristics of a defect. Smaller defects showed a higher densifcation rate, while the defect shape did not have signifcant efect on such rate.
The stress–strain behavior of ceramics, such as aluminum titanate, has certain features that are unusual for brittle materials—in particular, a substantial nonlinearity under uniaxial tension, and load–unload hysteresis caused by the sharp increase of the incremental stiffness at the beginning of unloading. These features are observed experimentally and are attributed to microcracking. Here we compare different degrees of stress–strain nonlinearity of aluminum titanate materials and quantitatively model them. We use advanced mechanical testing to observe the mechanical response at room and high temperature; electron microscopy, and X-ray refraction radiography to observe the microstructural changes. Experiments show that two types of microcracks can be distinguished: (i) microcracks induced by cooling from the sintering temperature (due to heterogeneity and anisotropy of thermal expansion), with typical sizes of the order of grain size, and (ii) much larger microcracks generated by the mechanical loading. The two microcrack types produce different effects on the stress–strain curves. Such microcracks and the features of the stress–strain behavior depend on the density of the cooling-induced microcracks and on the distribution of grain sizes. They are modeled analytically and numerically.