20005 research outputs found
Sort by
A multi-resolution MPS-PD coupling method for three-dimensional fluid-structure interaction involving fracture
No full textIn this paper, an in-house solver MPSPD-SJTU is developed to address 3D Fluid-Structure Interaction (FSI) problems with fracture. The Moving Particle Semi-implicit (MPS) method is utilized to simulate the violent free-surface flows while the PeriDynamics (PD) method is employed to model the structures with large deformations and fractures. Besides, a Kernel Function Based Interpolation Technique (KFBI) is incorporated into the coupling method to handle the MPS particles and PD particles with different resolutions. Firstly, the MPS solver is utilized to simulate dam break flow impacting a fixed obstacle, serving to evaluate its accuracy and reliability. Subsequently, the PD-based structural solver is validated by simulating an oscillating cantilever plate, a clamped-clamped elastic beam under uniformly distributed load, the Kalthoff-Winkler impact test, and three-point bending of an ice beam. Next, the MPS-PD model is validated via some benchmark tests, such as flood discharge with an elastic plate, hydrostatic water column on an elastic plate, dam-break flow with a hanging elastic plate, as well as dam-break flow with an elastic obstacle. The numerical results obtained in this study exhibit good agreement with experimental data as well as those from other numerical methods. Finally, the coupling model is applied to investigate the dam-break flow impacting a hanging elastic plate with a prefabricated crack
Torsional galloping with magnetic repulsion vertical axis current turbine
No full textThe article proposes an innovative concept of a current turbine based on the torsional galloping phenomenon, combined with the magnetic repulsion of a streamlined thin vertical plate, intended to harvest energy from low-head water currents. The driving proposal is to exploit the inherent instability of the arrangement. Hence, a careful non-linear instability analysis using the Poincaré method was performed numerically and experimentally. The mathematical model based on Bessel functions and potential theory was validated experimentally. The study carried out different configurations, comprising different stability conditions. The variances are selected by variance analysis. Since energy extraction is performed due to its unstable behavior, the occurrence of a stable condition must be avoided. It has been possible to select the condition where the turbine presents the most significant instability and the one that will provide higher efficiency. A final experimental set-up reached conclusive and successful feasibility
Towards in vivo MRI axon radius mapping: insights from MRI-scale histology and experimental validation
No full textAxons are micrometer-thin cables in the brain whose size is a potential biomarker for neurological disorders. While magnetic resonance imaging can, in theory, detect axon size, quantitative experimental proof has so far been lacking. We provide this proof via comparison with millions of axons from human brain microscopy and identify remaining challenges for clinical adoption.Axone sind mikrometerdünne Kabel im Gehirn, deren Größe ein möglicher Biomarker für neurologische Erkrankungen ist. Obwohl die Magnetresonanztomographie theoretisch die Axongröße erfassen kann, fehlte bislang ein quantitativer experimenteller Nachweis. Wir erbringen diesen Nachweis durch den Vergleich mit Millionen von Axonen aus dem menschlichen Gehirn und identifizieren noch zu überwindende Hürden für die klinische Anwendung
Surrogate modeling of brake squeal and brake praticle emissions from multivariate time-series data using deep learning
No full textμCT based quantification of pellet char morphology: Effects of biomass pelletization and fluidized bed pyrolysis
No full textSpruce wood pellets were produced with flat dies of different press-channel diameter-to-length ratios (1:3, 1:4, 1:5) and pyrolyzed at 900 °C for 4 min in a fluidized bed (FLB) and, for comparison, in a control setup (CS) where hot gas flowed around the pellets. The study includes (a) implementing a μCT radial porosity analysis to relate pellet-char structure to mechanical stability across distinct gas–solid contacting modes; (b) developing a μCT-based sand correction to separate entrained quartz from pellet char, reconciling image- and density-derived porosities; and (c) providing μCT evidence of fines enrichment toward the pellet core prior to pyrolysis, consistent with central-cavity formation under FLB conditions. FLB-pyrolysis yielded degraded pellet chars with pine cone-like morphology and large central cavities; μCT-resolved porosity increased by 6–12× relative to the wood pellets, depending on initial density. CS-pyrolysis produced chars that retained cylindrical shape and radial porosity distributions similar to untreated pellets, albeit at higher absolute porosity. The sand-mass correction indicated small fractions that minimally affected partial porosity but biased density-derived values. Across both conditions, extensive carbonization and loss of inter-particle bonding led to strength ranked 1:5 > 1:4 > 1:3, mirroring initial pellet quality
Maschinelles Lernen zur Überwachung semi-automatischer Bohrprozesse im Flugzeugbau
No full textDie große Anzahl semi-automatisch gefertigter Bohrungen bei der Luftfahrzeug-Strukturmontage stellt mit den heutigen Sensortechnologien eine vielversprechende Grundlage für die Anwendung des maschinellen Lernens (ML) beim datenbasieren Prozessmonitoring dar. In dieser Arbeit wird die ML-basierte Anomaliedetektion, Prozesszustand-, Werkstückqualität- und Werkzeugzustand-Überwachung beim semi-automatischen Bohren untersucht, um die erreichbaren Vorhersagegüten und optimalen Methoden der einzelnen Modellierungsschritte zu identifizieren.The large number of semi-automatically drilled holes in aircraft structural assembly provides a promising basis for the application of machine learning (ML) in data-based process monitoring using today's sensor technologies. This thesis investigates ML-based anomaly detection, process status, workpiece quality, and tool condition monitoring in semi-automatic drilling in order to identify the achievable prediction accuracies and optimal methods for the individual modeling steps.Bundeswirtschaftsministeriu
Modeling, sensing, and control for agile trajectory tracking of soft robots
No full textAccurate, agile trajectory tracking is important for modern soft robotic applications. This requires models that accurately represent dynamic behavior, soft sensors for state estimation, and dynamic controllers. In this work, first, different model-based and data-driven modeling approaches for soft robots are presented and evaluated in terms of accuracy and efficiency. Second, curvature sensors that can be integrated into the soft robot body to measure its configuration are presented and evaluated experimentally. Third, feedforward and feedback control approaches for soft robots are presented and evaluated in simulations and experiments.Für moderne Softroboteranwendungen ist genaues und agiles Trajektorientracking von großer Bedeutung. Dies erfordert genaue dynamische Modelle, weiche Sensoren zur Zustandsschätzung sowie dynamische Steuerungs- und Regelungsverfahren. In dieser Arbeit werden zunächst verschiedene modellbasierte und datengetriebene Modellierungsansätze für Softroboter vorgestellt und verglichen. Anschließend werden Krümmungssensoren präsentiert, die sich in den Softroboter integrieren lassen, um dessen Konfiguration zu messen. Abschließend werden Steuerungs- und Regelungsansätze für Softroboter präsentiert und in Simulationen sowie Experimenten evaluiert
Reactors for fluid-fluid reactions: Loop reactors
No full textVarious types of loop reactors are presented, which can be operated in single-, two-, or multiphase modes. The fluid dynamic laws for the design of the reactors are communicated. The focus is on the jet loop, which is frequently used in reaction technology. Of central importance, for example, for the residence time distribution or the generation of interfaces in gas/liquid systems, is the loop circulation speed, which is particularly emphasized. Considerations are also made for scale-up. The importance of loop reactors for the chemical industry is demonstrated using examples
Ensemble adaptive gated multi-fidelity neural network for Bayesian optimization: Application to hydrofoil design
No full textHigh-fidelity computational fluid dynamics (CFD) simulations provide critical predictive accuracy in marine and ocean engineering design; however, their substantial computational expense often renders direct optimization infeasible. To alleviate this limitation, surrogate models approximate expensive objective functions from a finite set of observations, thereby enabling more tractable design exploration and optimization. Our objective is to build a novel, general-purpose multi-fidelity surrogate modeling approach that integrates seamlessly into a Bayesian optimization framework and remains robust under sparse high-fidelity data. We propose an adaptive gated multi-fidelity neural network (AGMF-Net), which incorporates three specialized expert subnetworks—linear, nonlinear, and residual—combined through a deep Mixture-of-Experts gating network that dynamically adjusts their contributions based on the input. To improve predictive uncertainty estimation, we ensemble multiple independently initialized AGMF-Net instances and use the resulting variance to guide sampling decisions. We embed this surrogate into a Bayesian optimization workflow driven by the logarithmic expected improvement acquisition function, which balances exploration and exploitation while maintaining numerical stability. We evaluated the proposed method against co-Kriging and the multi-fidelity neural network baseline on benchmark functions. AGMF-Net achieved higher initial predictive accuracy, rapidly converged to global optima, and maintained lower mean absolute relative error during optimization iterations. Finally, we applied the framework to a hydrofoil design optimization. The model successfully identified a subtle camber modification that improved the lift-to-drag ratio by 41.6 % compared to the baseline geometry, demonstrating that AGMF-Net can accelerate CFD-driven hydrodynamic design scenarios that combine sparse high-fidelity data with cheaper simulations. These results highlight the potential of adaptive gating and ensemble uncertainty quantification to accelerate design exploration and improve solution quality when only limited high-fidelity evaluations are feasible
Efficient massively space-time-parallel simulations with adaptive spectral deferred correction
No full textSpectral Deferred Correction (SDC) is a method for numerically integrating initial value problems. The method iteratively generates solutions to fully implicit Runge-Kutta methods with forward substitution using low order solves. This allows great flexibility, for instance in terms of splitting techniques or inexact solves. Furthermore, various parallel-in-time extensions exist that parallelize the solution of a single time-step or solve multiple steps concurrently.
We propose two adaptive step size selection algorithms that tailor the ideas behind embedded Runge-Kutta methods to SDC. Both are completely generic and work only on intermediate values within the time-integration process. We show, with a range of experiments, that computational efficiency can be boosted significantly by employing these algorithms compared to standard SDC. Furthermore, we show that parallel-in-time adaptive SDC is competitive with state-of-the-art Runge-Kutta methods for stiff partial differential equations.
We also show that adaptivity increases the resilience against soft faults in SDC. Soft faults are unanticipated alterations of the data stored in memory, brought about, for instance, by environmental radiation. Iterative or adaptive methods inherently provide an elevated level of resilience, which is well known also in the context of the embedded Runge-Kutta methods that the adaptive step size selection is based on.
We then move on to port implementations of partial differential equations within the prototyping library pySDC to GPUs and make extensive space-time-parallel scaling tests. We find that the parallel-in-time extension diagonal SDC can help extend the scaling capabilities and allowed to run a Gray-Scott example on 3584 GPUs at decent parallel efficiency. Finally, we demonstrate that findings from the previous experiments translate to practical use via space-time-parallel production runs of Gray-Scott and Rayleigh-Benard convection using adaptive SDC.Spectral deferred correction (SDC) ist ein Zeitschrittverfahren bei dem Lösungen zu voll-impliziten Runge-Kutta (RK) Verfahren iterativ gewonnen werden. Wir entwickeln zwei Methoden zur adaptiven Schrittweitensteuerung und demonstrieren deren Potenzial zur Effizienzsteigerung an vier nichtlinearen Problemen. Dann messen wir Beschleunigung durch Parallelisierung mittels diagonalem SDC und Block Gauß-Seidel SDC, wobei Zeit-paralleles SDC für steife PDEs schneller war als RK Verfahren. Danach bestätigen wir experimentell, dass adaptives SDC gegen spontane Bit-Flips schützen kann, welche auf großen Supercomputern regelmäßig zu erwarten sind. Schließlich entwickeln wir Raum-Zeit-parallele Implementierungen von Spektralverfahren und diagonalem SDC, die für Rayleigh-Benard Konvektion bis tausenden von CPUs und für das Gray-Scott Modell bis tausenden von GPUs skalieren