18069 research outputs found
Sort by
Calculating the wing lift distribution with the Diederich method in Microsoft Excel
Full text linkAim of this project is to provide the Diederich Method for calculating the lift distribution of a wing in a Microsoft Excel spreadsheet based on didactic considerations. The Diederich Method is described based on primary and secondary literature. Diagrams are digitized so that the method can run automatically. To optimize the lift distribution of the wing, the elliptical and triangular lift distribution as well as Mason's lift distribution are offered for comparison. A method for calculating the maximum lift coefficient of the wing is integrated into the Diederich Method. To do this, the maximum lift coefficients of the airfoils at the wing root and at the wing tip must be entered into the program. The calculation assumes a trapezoidal wing. Both wing sweep and linear wing twist can be taken into account. The aspect ratio must not assume values that are too small. Subsonic flow and unseparated flow are assumed. Since only the wing is described, all other influences such as from the fuselage or from the engines are not taken into account. The Excel workbook was created for teaching in aircraft preliminary design. At the moment, the Diederich Method is apparently nowhere offered as a spreadsheet. With this work, this gap can be closed.NonPeerReviewe
Modelling loads on closely-spaced rotors
Full text linkMulti-rotor turbines offer the possibility of exploiting local blockage effects to improve energy yield through deploying rotors in close proximity to each other. A consequence of closely spacing rotors in such systems is increased mean and transient aerodynamic loading on the blades compared to single rotor systems. While variations in blade loading due to passing neighbouring rotors are observed in experiments and numerical simulations, mid-fidelity actuator models, such as the actuator line, may require bespoke sub-models and corrections for application to multi-rotor systems.NonPeerReviewe
Rotor–rotor interaction in multi-rotor arrays
Full text linkMulti-rotor wind turbines promise lighter structures and higher capacity, yet operating experience remains scarce. This talk reviews rotor–rotor interaction across three scales—blade-to-blade, rotor-to-rotor and array flow—drawing on lifting-line vortex, actuator-line RANS (Reynolds-averaged Navier–Stokes)/LES (Large eddy simulation) and actuator-disk CFD (Computational fluid dynamics) studies. Focus is on aerodynamic interactions shaping performance and load distributions, with brief reference to implications for structures, fatigue and control. Consistent findings, open questions, and future directions for aerodynamic research on multi-rotor arrays are outlined.NonPeerReviewe
Virtual environment and automated physical rolling maze as experimental platform for deep reinforcement learning
No full textIn the context of training competent future engineers, we develop platforms that shall help students to build practical competencies by working on challenging tasks for creative and highly motivating applications. Several of these platforms use systems that autonomously learn to master control tasks. Such systems are typically based on deep reinforcement learning (DRL), and related algorithms are frequently demonstrated by agents that learn to play games. In the following, we report on first results related to a platform where AI agents learn to manoeuvre balls through virtual and physical mazes while avoiding dropping into holes.NonPeerReviewe
Entwicklung eines QGIS-Plugins zur Vorplanung von Nahwärmenetzen basierend auf einem Biased Random-Key Genetic Algorithm
Full text linkIn dieser Arbeit wird ein Software-Tool als QGIS-Plugin entwickelt, um Gebäudegruppierungen zu identifizieren, für die die Neuerstellung eines Nahwärmenetzes effizient wäre. Dazu wird eine Datenverarbeitungspipeline modelliert, die frei verfügbare geographische Daten und graphentheoretische Algorithmen nutzt, um ein dreischrittiges, heuristisches Clusteringverfahren durchzuführen. Die zentrale Problemstellung besteht in der Zuordnung von Gebäuden mit Wärmebedarf zu Wärmequellen begrenzter Leistung. Dies wird als kapazitäts-beschränktes Clusteringproblem formuliert und aufgrund seiner Komplexität mit einem biased random-key genetic algorithm (BRKGA) gelöst. Im Rahmen der iterativen Optimierung werden die Kosten für Rohr- und
Grabenbau minimiert, um kostengünstige Gebäudegruppierungen miteinander zu verbinden. Das Tool ist speziell für kleine Wärmenetze mit maximalen Wärmeleistungen von ≤250 kW ausgelegt, deren Wärmequellen in zentralen Gebäuden der identifizierten Gruppierungen platziert werden könnten. In einem strukturierten Softwaredesignprozess werden Anforderungen definiert und ein Lösungsansatz umgesetzt. Die Software ermöglicht die Analyse verschiedener Rohrverlegungsstrategien, etwa entlang des Straßennetzes oder durch direkte Gebäudeverbindungen. Durch die Umsetzung der zentralen Anforderungen bietet sie eine flexible Grundlage für Weiterentwicklungen und Vergleiche, um Akteure bei den Herausforderungen der Wärmewende praktisch zu unterstützen.This thesis develops a QGIS-plugin to identify building clusters suitable for cost-efficient small-scale district heating networks. To reach this aim a data processing pipeline is designed, leveraging open geospatial data and utilizing graph theory algorithms to inform and conduct a three-step heuristic clustering process. The core challenge of this process is the assignment of buildings with a certain heating demand to heating sources with a limited heating power. As such it is formulated as a capacitated clustering problem and is solved by using a biased random-key genetic algorithm (BRKGA) due to the problem’s inherent complexity. Throughout the iterative optimization process, costs for pipes and trenching are minimized to identify cost-effective building groups for interconnection. The tool targets small networks (≤250 kW), enabling heat source placement in central buildings of identified clusters. A structured software design process translates requirements into a functional solution. The developed software enables the analysis of various pipe laying strategies, for example according to the street network or through direct building connections. By fulfilling the core requirements,
the plugin offers a basis for further development and comparisons and should therefore be able to aid stakeholders through the challenges of decarbonizing the heating supply
Entwicklung und Validierung eines prototypischen IDE-Plugins zur LLM-gestützten Erstellung einer Softwarearchitektur
Full text linkDer Entwurf von Softwarearchitektur erfordert spezialisiertes Wissen. Zudem scheinen spezialisierte LLM-Werkzeuge für den Architekturentwurf unzureichend in bestehende IDEs integriert zu sein.
Hier setzt der innovative Ansatz dieser Arbeit an: Ein LLM-Chatbot-basiertes IDE-Plugin zum Softwarearchitekturentwurf wird entwickelt und validiert. Dafür wird ein bestehendes Plugin für Attribute-Driven Design 3.0 spezialisiert und die Erfüllung definierter Kriterien durch erzeugte Architekturentwürfe an einem Fallbeispiel untersucht. Die Ergebnisse zeigen die Umsetzbarkeit des Plugins, das erfolgreich Architekturentwürfe erzeugen kann. Diese können vielen Architektureigenschaften entsprechen, aber Inkonsistenzen und Halluzinationen können auftreten.The design of software architecture requires specific knowledge. In addition, specialized LLM tools for architecture design seem to be insufficiently integrated into existing IDEs.
This work introduces an innovative approach: An LLM-chatbot-based IDE plugin for software architecture design is developed and validated. To this end, an existing plugin is specialized for Attribute-Driven Design 3.0 and the fulfillment of defined criteria by generated architecture designs is analyzed using a case study. The results show the feasibility of the plugin, which can successfully generate architecture designs. These can fulfill many architectural properties, but inconsistencies and hallucinations can occur
Impedance simulation with the P2D approach and frequency-domain sensitivity analysis of model parameters
No full textReviewPendin
Spatio-Temporal Shifts in Citizen Science Data: Detecting Disruptions in Bird Sightings with Change Point Analysis
Full text linkSpezies, einschließlich Vögel, können Verschiebungen in ihrer Verbreitung und Häufigkeit erfahren, was sich auf Ökosysteme und die Biodiversität auswirken kann. Change-Point-Detection (CPD)-Methoden sind wertvolle Werkzeuge zur Identifikation solcher Veränderungen. Citizen Science bietet hierfür großflächige Datensätze, bringt jedoch auch beobachterbedingte Verzerrungen mit sich. Dies wirft Fragen zur Verlässlichkeit etablierter CPD-Algorithmen bei der Anwendung auf solche Daten sowie zu ihrer Akzeptanz unter Fachexperten auf.
Diese Arbeit greift diese Problematik auf, indem sie einen CPD-Ansatz unter Verwendung des "Bayesian Estimation of Abrupt Change, Seasonality, and Trend" (BEAST)- Algorithmus auf einen Citizen-Science-Vogeldatensatz anwendet. Vor der BEASTAnalyse wird eine Preprocessing-Pipeline entwickelt, um Beobachterverzerrungen zu reduzieren. Die Evaluation untersucht die Genauigkeit von BEAST sowie dessen ökologische Relevanz im Kontext von Citizen Science. Detektierte Veränderungspunkte wurden quantitativ mit dokumentierten ökologischen Ereignissen validiert, während Ornithologen die ökologische Plausibilität und praktische Relevanz qualitativ bewerteten. Die Ergebnisse zeigen, dass BEAST ökologisch bedeutsame Veränderungspunkte zuverlässig erkennt, wobei seine Sensitivität von der Datenaggregation und den gewählten Preprocessing-Strategien abhängt. Obwohl CPD manuelle Bewertungen nicht ersetzt, wird es als wertvolle Ergänzung angesehen, um subtile oder unerwartete Veränderungen aufzudecken, die Echtzeit-Überwachung ökologischer Prozesse zu unterstützen und das Retraining von Machine-Learning-Modellen zu informieren. Trotz des spezifischen Anwendungsfalls, unterstreicht diese Studie das breitere Potenzial von CPD in der Citizen Science, indem sie zeigt, dass mit robustem Preprocessing und Expertenvalidierung zeitnahe ökologische Erkenntnisse gewonnen werden können.Species, including birds, can undergo sudden shifts in distribution and abundance due to environmental changes, human activities, or natural variability, which can impact ecosystems and biodiversity. Change Point Detection (CPD) methods are valuable for identifying these shifts. For this, citizen science offers large-scale datasets, but also introduces observer-related biases, raising questions about the reliability of established CPD
algorithms when applied to such data, and their trustworthiness among domain experts. This thesis addresses this concern by implementing a CPD approach using the Bayesian Estimation of Abrupt Change, Seasonality, and Trend (BEAST) algorithm on a citizen science bird dataset. Prior to BEAST analysis, a tailored preprocessing pipeline is developed to mitigate user bias. Evaluation examines BEAST’s accuracy and ecological relevance in citizen science contexts. Detected change points were quantitatively validated against documented ecological
events, while ornithologists qualitatively assessed ecological plausibility and practical relevance. Findings indicate that BEAST reliably detects ecologically meaningful change points, though its sensitivity depends on data aggregation and preprocessing strategies. While not replacing manual assessments, CPD is seen as a valuable complement to uncover subtle or unexpected changes, supporting real-time ecological monitoring, and informing machine learning model retraining. Though case-specific, this study underscores CPD’s broader potential in citizen science, enabling timely ecological insights when paired with robust preprocessing and expert validation
Digital transformation in higher education : artificial intelligence tools, pedagogical practice, and data literacy development
No full textThis paper investigates how AI tools influence data literacy and pedagogical practices in global higher education contexts. It presents findings from a global workshop that examined the role of AI in academic instruction, synthesizing empirical insights from educators and researchers in China, South Africa, the United States, and Europe. Through an explorative qualitative approach, this study clusters expert insights into key areas: AI-enhanced pedagogical frameworks, good practice examples, student engagement strategies, and ethical considerations. The findings emphasize concrete applications of AI in curriculum development, institutional challenges, and the necessary policy frameworks for the adoption of responsible AI in higher education.PeerReviewe
Stress-based optimization of components and supports for sinter-based additive manufacturing
No full textSinter-based additive manufacturing (SBAM) processes, such as Cold Metal Fusion (CMF), combine the geometric freedom of additive manufacturing with the scalability of powder metallurgy, but part distortion and collapse during debinding and sintering remain critical design challenges. This study presents a revised stress-based optimization framework to address these issues by integrating sintering-specific load cases into topology optimization. In contrast to earlier approaches, the revised workflow applies all load cases to the upscaled green-part geometry. This adjustment mitigates the non-linear scaling effects of dead load-induced stresses. A Case study, including a steering bracket for a Formula Student racing car, demonstrates that the revised method improves not only sinterability but also application-related performance compared to earlier approaches. In addition, a semi-automated procedure for generating sinter supports is introduced, allowing stable processing of geometries without planar bearing surfaces. Experimental validation confirms that optimized supports effectively prevent part failure during post-processing, though challenges remain in separating complex freeform geometries. Finally, the influence of stiffness on sintering-induced deformations is investigated, showing that higher stiffness configurations significantly reduce dimensional errors. Together, these results highlight stress- and stiffness-based optimization as tools to enhance the reliability, efficiency, and design freedom of SBAM.PeerReviewe