13979 research outputs found
Sort by
Optimization of Mechanical and Electrical Properties in Age-Hardened Aluminum Alloy through HPTE Processing
No full textAluminum alloys are commonly used due to their versatility and tunability of properties through addition of solutes and heat treatments. This study focuses on age-hardenable Al-Mg-Si alloys, which are lightweight and extensively used in the electrical industry. Precipitate hardening is the way to improve strength in this alloy system, where the size of precipitates depends on temperature and aging time.
However, the presence of solute atoms is problematic as it degrades electrical conductivity (EC) by reducing the mean free path of conducting electrons. This inherent trade-off between strength and electrical conductivity presents significant challenges for industrial applications, where both properties are crucial for optimal overall performance.
This bottleneck serves as the primary motivation for this study, focusing on simultaneous improvement of strength and EC. In this study, we introduce an approach that combines severe plastic deformation with subsequent heat treatment to optimize both strength and conductivity. Despite the proven efficacy of severe plastic deformation in strengthening materials, previous studies in this direction
have predominantly operated in a limited scope of laboratory scale. Our research aims to bridge this gap by exploring the application of high-pressure torsion extrusion (HPTE) as a viable method for implementing these techniques at a larger scale. Mechanical properties were evaluated by microhardness and tensile testing. The results showed a remarkable improvement in strength while EC remained unchanged. Furthermore, modeling of strength and EC, based on microstructural
characterization, revealed good agreement with experimental values.
The second objective is to conduct an atomic-level investigation of the precipitates formed under conditions of extreme deformation and subsequent annealing. Advanced microscopy and EDX microanalysis demonstrated that the material processing led to the development of various hardening phases with distinct morphologies and compositions. The high density of defects induced by HPTE deformation resulted in a predominantly disordered structure for most particles. The particles exhibit a set of orientation relations with the Al matrix, which differ from those established for the studied alloy.
Furthermore, a considerable number of nanometer-sized precipitates exhibit a core-shell structure or evidence of simultaneous co-precipitation of several secondary phase particles. These observations lead to the conclusion that nucleation and growth of hardening precipitates occur in non-equilibrium conditions
Risk profiles for self-medication with analgesics among elite German handball players
No full textThis study presents empirical evidence on self-medication with analgesics among elite German handball players. The aim is to elucidate the intentions behind therapeutic and preventive analgesic use and identify high-risk profiles for this complex health-risk behavior based on psychosocial factors. Data were collected from 459 handball players (233 female, 226 male athletes) across the highest German divisions (1st – 3rd and German Youth Handball Bundesliga) through a quantitative online survey conducted between October 2021 and April 2022. The survey requested demographic information and details on analgesic use patterns, intentions behind analgesic use, and psychosocial variables. The study shows that athletes use self-medication with painkillers as a coping strategy to manage distractions from acute injuries and pain as well as to prevent expected distractions. Engaging in self-medication allows athletes to navigate the pressures of elite sport. Classification tree analysis revealed distinct risk profiles for therapeutic and preventive analgesic use. Key factors of self-medication include age, athletic identity, willingness to compete hurt, and perceived performance state. The high-risk group for therapeutic self-medication with analgesics consists of senior athletes. The high-risk group for using analgesics preventively, in contrast, is composed of athletes over the age of 17 with strong negative affectivity, a heightened willingness to take risks, and good perceived performance. Our findings highlight the necessity for specific interventions to address the multifaceted motivations behind self-medication and promote safer practices in elite sports. Future research should concentrate on longitudinal studies and diverse athlete populations and use qualitative studies to gain a deeper understanding of the determinants of analgesic use and to develop effective intervention strategies
Physically Interpretable Machine Learning for Lithium-Ion Batteries
No full textLithium-ion batteries, with their high energy and power densities, are key for decarbonization. Optimizing battery design and operation is important for reducing costs, resource usage, and energy footprints. However, battery degradation is complicated, path-dependent, and not fully understood. Measurements can be noisy, and many important parameters cannot be directly measured in commercial battery cells. This thesis addresses these challenges using physically interpretable machine learning across three interconnected problems, from individual cells to battery systems.
First, we investigate the interpretability of high-dimensional regression for battery life prediction based on cycling test data, which is important to accelerate battery testing and development and to improve the understanding of degradation. High-dimensional regression can directly work with high-dimensional measurement data, yielding linear coefficients. We showcase interpretability challenges in this context and develop a method to understand how the nullspace and regularization shape regression coefficients. The nullspace of the data matrix allows very different regression coefficients to yield identical predictions. We demonstrate that the fused lasso, a regression method that yields coefficients not orthogonal to the nullspace, improves physical interpretability on two battery aging data sets. The insights gained help make informed design choices for building regression models on high-dimensional data and reasoning about potential underlying linear models, which is generally important for system optimization and improving scientific understanding.
Another important challenge this thesis addresses is diagnosing batteries with electrochemical impedance spectroscopy. Experts typically match impedance spectra with equivalent circuit models to track battery performance. Here, we show three approaches to automate the classification of equivalent circuit models by leveraging a large synthetic impedance data set. The results build on the BatteryDEV hackathon that was organized during this dissertation. The best-performing approach is a gradient-boosted tree model in combination with automatic feature generation. Interpretation shows that the most important features align with physically meaningful impedance characteristics of specific circuits. The next best model is a random forest model, which uses the impedance data directly. The convolutional neural network trained on Nyquist plot images achieves a lower classification accuracy. The classification results highlight the potential of automatic equivalent circuit model selection for impedance data.
Finally, this thesis develops online health monitoring systems for fault detection using data from real-world battery operations of lithium-iron-phosphate battery systems. The data set contains 28 systems returned to the manufacturer for warranty, each with eight cells in series, totaling 224 cells and 133 million data rows. We use recursive spatiotemporal Gaussian process models to separate the time-dependent and operating-point-dependent resistances. These processes scale linearly with the number of data points, allowing online monitoring. We develop probabilistic fault detection rules. Often, only a single cell shows abnormal behavior or a knee point, consistent with weakest-link failure for cells connected in series, amplified by local resistive heating. The results further the understanding of how battery systems degrade and fail in the field and demonstrate how online monitoring and early fault detection improve battery safety.
The physically interpretable models developed in this thesis support research and development, further the understanding of battery degradation, and improve safety. This thesis made methodological contributions and generated scientific insights by applying existing methods to battery data sets. We published the synthetic impedance data set and the field data set. Furthermore, we open-sourced four software repositories for reproducibility and reusability. The approaches developed in this thesis are flexible and are expected to be adaptable to novel materials and systems
Digitale Geländemodelle in der Auenforschung: Analyse natürlicher Dynamik und anthropogener Eingriffe
No full textDigitale Geländemodelle (DGMs) eröffnen uns eine faszinierende Perspektive auf die Entwicklung der mitteleuropäischen Auenlandschaften. So speichert die Auentopographie Informationen von Gewässerverläufen und flussbegleitender Oberflächenformen, wie die Schleifen alter Flussmäander, Uferwälle oder auch Gleithangschichtungen. Damit erzählen sie von der ununterbrochenen Dynamik der Flüsse, deren Lauf sich durch seitwärts gerichtete Verlagerung immer wieder veränderte. Selbst die trägen, verzweigten Wasserwege mancher Tieflandflüsse, die sich bei geringer Fließgeschwindigkeit wie Adern in der Landschaft verzweigen, finden im Relief eine beeindruckende Präzision. Doch nicht nur die Natur, auch der Mensch hat seine Spuren hinterlassen: Begradigungen, Durchstiche und Kanäle, die für Mühlen, Flößerei oder Hochwasserschutz angelegt wurden, sind in der Topographie als klare Linien und Brüche in den Auenlandschaften erkennbar. Jede dieser Modifikationen fügt dem Relief eine Schicht an Geschichte hinzu. DGMs machen die Topographie der Landschaft flächenhaft und quantitativ erforschbar. Damit lassen sich nicht nur frühere Flussverläufe rekonstruieren, sondern auch die Dynamik und räumliche Ausdehnung von Veränderungen quantitativ analysieren. DGMs sind somit ein Schlüsselwerkzeug, um die komplexen Wechselwirkungen zwischen natürlicher Flussdynamik und menschlichem Einfluss zu entschlüsseln
Computational Analysis of Literary Communities: Event-Based Social Network Study of St. Petersburg 1999-2019
Full text linkThis paper presents a computational analysis of literary networks in St Petersburg from 1999 to 2019, using data from the SPbLitGuide newsletter and exploring cultural connections through event co-participation. By processing 15,012 cultural events with 11,777 participants in 862 venues, we reveal the structure and evolution of the literary network in post-Soviet Russia. Our methodology combines network, spatial, and temporal approaches, demonstrating how systematic event recording can capture patterns of literary community formation typically invisible to traditional literary history. The study covers the last decades of St Petersburg's predominantly offline literary life before its radical transformation in the post-2020/2022 period, providing both a historical record and a methodological framework applicable to other cultural contexts. Our findings show a complex ecosystem characterised by dense local clusters, influential bridge figures, and distinct community boundaries, while documenting crucial shifts in the city's literary infrastructure over two decades
The Outward Turn: Geocoding the Expansion of Fictional Space in Russian 19th Century Literature
No full textWe examine the large-scale geospatial dynamics of Russian prose literature
in the 19th century. Specifically, we analyze how the distribution of location
mentions shifts from the early 19th-century romantic era to the late 19th-century
realist period. We demonstrate how realist literature, with its emphasis on
portraying ’typical characters in typical settings’, moves away from the historical
(and often heavily mythologized) landscapes of Russia, Poland, Ukraine, and the
Baltics. Instead, it increasingly focuses on the then-new capital, Saint Petersburg,
as well as Western Europe and the expanding eastern and southern peripheries
of Russia, reflecting the country’s ongoing military and economic expansion
Structure–Function Relationship of the Most Abundant Ceramide Subspecies Studied on Monolayer Models Using GIXD and Langmuir Isotherms
No full textThe main lipid compounds of the outermost layer of human skin are ceramides (CERs), free fatty acids, and cholesterol. Although numerous studies performed in the past could demonstrate the importance of these lipids for an intact skin barrier function, knowledge about the impact of each single component on the lamellar lipid films is still lacking. Especially, the CERs are a very heterogeneous group with high relevance for a proper barrier. It was found that the reason for the high stability of the lamellae is related to the lipid structure and function, with the type and extent of interactions between the head groups of the individual CER subspecies being particularly important. Elucidating these at the molecular level could help us to understand CER phase behavior in general. Using grazing incidence X-ray diffraction and measurements of Langmuir isotherms, the current work investigated the lateral packing of the monolayers of different subclasses of C18:0 CERs at air–water interfaces, including phytosphingosine, sphingosine, and dihydrosphingosine CERs, all with either α-hydroxy and nonhydroxy N-acylated fatty acyl. We were able to observe clear effects of the minimal differences in the polar headgroup structures of the sphingoid bases, with respect to the number and position of hydroxyl groups and double bonds, on the CER arrangement regarding the compressibility and structure of the films they formed, revealing that the hydroxyl group at the C4 of the phytosphingosine CERs leads not only to the formation of a hydrogen bond network but also to a stable suprastructure, which might be of high benefit for the barrier properties of intact skin
Pore scale reactive flow simulation in catalytic particulate filters based on high resolution X-ray CT images
No full textThis study presents a comprehensive pore-scale simulation of reactive flows in particulate filters, focusing on selective catalytic reduction diesel particulate filters (SDPF). Emphasis is placed on the impact of internal washcoat pores and the development of a zero-dimensional surrogate model for first-order reactions. Nano-computed tomography (nano-CT) with a voxel resolution of 364 nm enabled detailed imaging and reconstruction of real filter samples, including the washcoat’s internal pores for the first time. Segmentation into pore, washcoat, and substrate phases provided a solid foundation for numerical simulations. To reduce computational cost, a representative elementary volume (REV) and a downsampling strategy were implemented and validated through geometric and flow parameter comparisons. Essential microstructural characteristics were preserved while significantly lowering computational demands. Results highlight the critical role of internal washcoat pores in mass transport and catalytic performance. SDPF samples with resolved internal pores outperformed those with closed washcoat structures (SDPF CWC), even with lower washcoat loading, underlining the importance of geometric connectivity and bimodal pore networks. Comparisons with catalytic gasoline particulate filters (cGPF) revealed the absence of internal washcoat pores in cGPF samples, suggesting finer pore structures. In the absence of experimental diffusion data, this work provides practical tortuosity values for estimating effective diffusion. To accelerate performance prediction, a zero-dimensional surrogate model based on residence time distribution (RTD) and washcoat efficiency was developed. Using Thiele modulus theory, it estimates conversion from effective diffusion lengths and efficiency factors, showing high accuracy across different geometries. The model emphasizes the equally important roles of RTD and washcoat efficiency in catalysis. Additionally, nonlinear (inhibition-controlled) reactions were simulated, demonstrating that spatial concentration profiles remain consistent across different inhibition factors at similar conversion levels. This finding enables simplification and generalization of spatial distributions in models of complex reaction kinetics
The role of Co–Ga₂O₃ interfaces in methane dry reforming
No full textAs the combination of Co with other non-noble metals is a viable way to improve the catalytic properties of Co in methane dry reforming (DRM), we studied an impregnated Co₃O₄/β-Ga₂O₃ powder catalyst to understand the influence of Ga and the catalytic role of the Co–Ga₂O₃ interface and the intermetallic compound CoGa in DRM. Co₃O₄/β-Ga₂O₃ undergoes a series of structural transformations during activation by reduction in hydrogen and under DRM conditions. Contact to the CO₂/CH₄ mixture without hydrogen pre-reduction yields CoGa₂O₄ spinel particles encrusting β-Ga₂O₃ without significant DRM activity. Hydrogen reduction transforms Co₃O₄/β-Ga₂O₃ initially to α-Co/β-Ga₂O₃, before it induces reactive metal–support interaction leading to the formation of bimetallic CoGa particles on β-Ga₂O₃. Subsequent improved DRM activity can be correlated to the decomposition of the intermetallic compound CoGa: according to operando X-ray diffraction CoGa re-transforms into α-Co/β-Ga₂O₃ during DRM. Hydrogen pre-reduction is a prerequisite for high DRM activity on Co₃O₄/β-Ga₂O₃, where intermediarily formed CoGa is decomposed under reaction conditions yielding a pronounced increase in the activity rivalling established noble metal and non-noble metal catalysts. A particular advantage of β-Ga₂O₃ is the suppression of coking and Co deactivation, as observed on a Ga-free Co/SiO₂ catalyst
Proteasen und Substrate für einen kompetitiven Fibrinogenschnelltest
No full textEin akuter Fibrinogenmangel stellt bei schweren Blutungen infolge von chirurgischen Eingriffen oder Traumata eine erhebliche Gefahr für Patienten dar. Die frühzeitige und präzise Bestimmung der Fibrinogenkonzentration im Plasma ist daher entscheidend für eine gezielte Substitutionstherapie mit Fibrinogenkonzentraten. Derzeit verfügbare Testverfahren sind jedoch mit hohen Kosten verbunden und weisen insbesondere im diagnostisch relevanten Konzentrationsbereich eine unzureichende Genauigkeit auf. Vor diesem Hintergrund wurde im Rahmen dieser Arbeit ein neuartiger enzymatischer Zwei-Felder-Kompetitionsassay entwickelt, der eine schnelle und kostengünstige Alternative darstellen soll.
Das grundlegende Prinzip des Assays basiert auf der kompetitiven Umsetzung eines fluorogenen Substrats in zwei Reaktionskompartimenten. Dabei wird das Enzymsubstrat auf einer Seite in einer Konzentration nahe der Michaelis-Konstante (KM) und auf der anderen Seite in Überschuss eingesetzt. Die Anwesenheit von Fibrinogen als kompetitiver Inhibitor in beiden Kompartimenten führt zu messbaren Unterschieden in der Umsatzgeschwindigkeit des Substrats, woraus ein Umsatzfaktor berechnet wird, der mit der Fibrinogenkonzentration korreliert.
Zu Beginn der Arbeiten wurde Thrombin als Testenzym verwendet und mit dem Substrat Z-GPR-AMC getestet. Die photometrische Auswertung wurde durch die Trübung infolge der Fibringerinnung gestört. Zwar konnte die Trübung durch den Zusatz des aggregationshemmenden Tetrapeptids GPRP vermindert werden, dies führte jedoch gleichzeitig zu einer Veränderung der Enzymkinetik. Fluoreszenzspektrometrische Messungen zeigten zwar verbesserte Ergebnisse, doch war die Steigung der Umsatzfaktorkurve im relevanten Konzentrationsbereich von 0–2 mg/mL Fibrinogen zu gering, um eine präzise Bestimmung zu ermöglichen.
Zur Verbesserung des Assays wurden daher alternative Enzym-Substrat-Systeme getestet. Basierend auf Literaturdaten wurden sogenannte Snake Venom Thrombin-like Enzymes (SVTLEs), insbesondere Ancrod und Batroxobin, in die Untersuchungen einbezogen. Diese Enzyme zeichnen sich durch eine hohe Affinität zu Fibrinogen aus. Während Ancrod keine zufriedenstellenden Ergebnisse lieferte, zeigte Batroxobin vielversprechende Eigenschaften, wenngleich die in der Literatur beschriebenen submikromolaren KM-Werte für Fibrinogen nicht erreicht wurden.
Ein zentrales Ziel der Arbeit war daher die Etablierung eines rekombinanten Expressionssystems für Batroxobin in Pichia pastoris. Sowohl ein Wildtyp-Gen als auch eine codonoptimierte Variante wurden in das pPIC9-Vektorplasmid kloniert und erfolgreich in den Wirtsstamm GS115 transformiert. Trotz vollständiger Integration des Gens in das Genom konnte zunächst kein Protein im Expressionsüberstand nachgewiesen werden. Erst durch zusätzliche Aktivitätstests auf Fibrinogen-Agaroseplatten sowie mit einem neu entwickelten Mikrotiterplatten-Assay konnte die Aktivität des exprimierten Batroxobins nachgewiesen und quantifiziert werden. Die Reinigung erfolgte mittels Nickelionen-Affinitätschromatografie. Trotz relativ niedriger Ausbeuten im Vergleich zu Literaturangaben konnte die Aktivität des rekombinanten Enzyms bestätigt werden. Weitere Optimierungen im Produktionsprozess, etwa durch Änderung der Kulturbedingungen und der Vorreinigungsstrategien, führten zu einer verbesserten Proteinqualität.
Ein weiterer zentraler Aspekt der Arbeit war die Entwicklung spezifischer FRET-Substrate für Batroxobin, da handelsübliche Substrate keine ausreichenden Umsatzraten zeigten. Die synthetisierten FRET-Substrate basierten auf der Thrombin-Spaltstelle des Fibrinogen Aα-Peptids und wurden mittels Festphasensynthese hergestellt. Erste Versuche ergaben, dass bei den für den Zweikompartimente-Test erforderlichen Konzentrationen ein erheblicher Teil der Fluoreszenz durch den sogenannten Inner-Filter-Effekt (IFE) absorbiert wurde. Durch Verwendung von Mikroküvetten und Reduktion der Substratkonzentration konnte der IFE deutlich reduziert werden. Die optimierten Substrate, insbesondere das verlängerte Peptid Aminobenzoyl-DFLAEGGGVRGPR-3-Nitro-Tyrosin, zeigten deutlich verbesserte Anfangsgeschwindigkeiten und niedrigere KM-Werte. Eine weitere Verlängerung der Sequenz mit Glycin- und Argininresten am N-Terminus führte zu einer zusätzlichen Steigerung der Substratumsetzung. Das längste Substrat, Abz-DFLAEGGGVRGPR-3-Nitro-Tyr, wies schließlich mit 339 µM die niedrigste KM und mit 60 nmol min-1 BU-1 die zweithöchste vmax auf.
Die abschließenden Tests mit dem optimierten Assaysystem und den eigens entwickelten FRET-Substraten zeigten, dass sich mit Batroxobin ein deutlich breiterer dynamischer Bereich der Umsatzfaktoren erreichen lässt als mit Thrombin. Die berechneten Z-Faktoren lagen über dem statistisch geforderten Schwellenwert von 0,7, was die Eignung des Assays für präzise Messungen im diagnostisch relevanten Bereich bestätigt.
Insgesamt konnten im Rahmen dieser Arbeit mehrere wesentliche Fortschritte erzielt werden. Batroxobin wurde als geeignetes Enzym für den Zwei-Felder-Kompetitionsassay identifiziert und erfolgreich rekombinant in Pichia pastoris exprimiert, auch wenn die Produktionsausbeute noch optimiert werden muss. Zudem wurden spezifische FRET-Substrate entwickelt und getestet, die eine zuverlässige Umsetzung durch Batroxobin ermöglichen. Mit diesen Komponenten konnte ein funktionsfähiges Assayverfahren zur Bestimmung der Fibrinogenkonzentration etabliert werden. Diese Ergebnisse schaffen eine belastbare Grundlage für die Weiterentwicklung zu einem praxistauglichen diagnostischen Schnelltests, der in Zukunft eine raschere und verlässlichere Therapieentscheidung bei Fibrinogenmangel unterstützen könnte