Politecnio die Bari - Catalogo di prodotti della Ricerca
Not a member yet
36616 research outputs found
Sort by
Physics informed neural networks for learning the horizon size in bond-based peridynamic models
No full textThis paper broaches the peridynamic inverse problem of determining the horizon size of the kernel function in a one-dimensional model of a linear microelastic material. We explore different kernel functions, including V-shaped, distributed, and tent kernels. The paper presents numerical experiments using PINNs to learn the horizon parameter for problems in one and two spatial dimensions. The results demonstrate the effectiveness of PINNs in solving the peridynamic inverse problem, even in the presence of challenging kernel functions. We observe and prove a one-sided convergence behavior of the Stochastic Gradient Descent method towards a global minimum of the loss function, suggesting that the true value of the horizon parameter is an unstable equilibrium point for the PINN's gradient flow dynamics
Laser Hybrid Welding of WAAM-Fabricated Dissimilar Steels: Microstructural Evolution and Mechanical Performance of AISI 308L and ER70S-6 Joints with 316L Filler
No full textIn this study, dissimilar stainless steel (AISI 308L stainless steel) and low-alloy steel (ER70S-6) plates fabricated via Wire Arc Additive Manufacturing were joined using Laser Hybrid Welding with AISI 316L stainless steel filler wire. Four welding parameter sets were tested by varying laser power (1000-1500 W) and Laser scanning speed (1000-1500 mm/min). Microstructural analysis using Scanning Electron Microscopy, Electron Backscatter Diffraction, and x-ray Diffraction revealed significant phase transformations in the weld zone, including the formation of δ-ferrite and γ-austenite, with retained δ-ferrite decreasing at higher laser powers. The welds exhibited refined grain structures, especially at the fusion boundaries and heat-affected zone, with Electron Backscatter Diffraction measured grain sizes of 5.13 ± 1 μm (ER70S-6 heat-affected zone), 16.33 ± 5 μm (316L weld), and 10.09 ± 6 μm (weld center). Maximum microhardness values reached 303 HV in the heat-affected zone of 308L, 297 HV in ER70S-6, and 245 HV in the weld center. Tensile testing showed a 43% increase in strength for the sample welded at 1500 W and 1000 mm/min compared to lower heat input conditions, confirming improved ductility and joint integrity. These findings demonstrate the feasibility and effectiveness of Laser Hybrid Welding for producing robust joints between wire arc additive manufacturing-fabricated dissimilar steels with enhanced mechanical and microstructural performance
Towards a digital twin of extrusion-based additive manufacturing: an experimentally validated numerical model of ironing process
No full textA numerical framework was developed to optimize ironing post-processing parameters to reduce porosity and manufacturing time while enhancing shape accuracy and ultimate tensile strength (UTS) of 3D printed polylactic acid (PLA) parts. The material extrusion (MEX) additive manufacturing of a top layer consisting of three adjacent strands was simulated using multiphase non-Newtonian computational fluid dynamics (CFD) under different values of extrusion temperature, layer height, and strand overlap. Simulations revealed that a 50% strand overlap prevented porosity, while a 25% overlap led to significant porosity, particularly at lower extrusion temperatures and higher layer heights. In this regard, ironing was identified as an effective post-processing technique. Therefore, numerical and experimental investigations have been done with respect to 25% strand overlap: ironing line spacing (LS) significantly influences porosity, while ironing speed (IS) affects manufacturing time. Higher ironing LS and IS yielded the best results at low layer heights, while a balanced approach optimized ironing at higher layer heights, resulting in a reduction up to 31.97% in overall manufacturing time if compared with traditional approach (i.e., lower values of ironing LS and IS). UTS improved by up to 21.49%, underscoring ironing impact on the final properties of PLA parts. The accuracy of the proposed CFD model is expected to lay the foundation for a reliable digital twin of MEX, improving part quality and reducing manufacturing costs and waste
A Comparison Study in Two PMaSynRels with Different Sintered Ferrite Shapes
No full textConventional permanent magnet assisted synchronous reluctance motors (PMaSynRels) are characterized by flux barriers partially filled with rectangular sintered ferrite, resulting in barrier shapes constrained to various rectangular combinations. For PMaSynRels with bonded ferrite, flux barriers could be completed filled, but the inferior magnetic properties of bonded ferrite restrict the performance of these PMaSynRels. Thus in the paper, two PMaSynRels with non-rectangular sintered ferrite are investigated and compared to explore the maximum torque potential. Through parameterization and optimization, the final two PMaSynRels are evaluated on torque, power factor, and efficiency across diverse operating conditions. The results prove that PMaSynRel with fluid-shaped sintered ferrite has lower torque ripple than PMaSynRel with circularshaped sintered ferrite. In terms of power factor and efficiency, the two PMaSynRels do not differ significantly
Diverse dynamics in interacting vortices systems through tunable conservative and non-conservative coupling strengths
No full textMagnetic vortices are highly tunable, nonlinear systems with ideal properties for being applied in spin wave emission, data storage, and neuromorphic computing. However, their technological application is impaired by a limited understanding of non-conservative forces, that results in the open challenge of attaining precise control over vortex dynamics in coupled vortex systems. Here, we present an analytical model for the gyrotropic dynamics of coupled magnetic vortices within nano-pillar structures, revealing how conservative and non-conservative forces dictate their complex behavior. Validated by micromagnetic simulations, our model accurately predicts dynamic states, controllable through external current and magnetic field adjustments. The experimental verification in a fabricated nano-pillar device aligns with our predictions, and it showcases the system's adaptability in dynamical coupling. The unique dynamical states, combined with the system's tunability and inherent memory, make it an exemplary foundation for reservoir computing. This positions our discovery at the forefront of utilizing magnetic vortex dynamics for innovative computing solutions, marking a leap towards efficient data processing technologies
Evaluating and Mitigating Terrorist Risk in Historical Outdoor Open Areas Including User Factors
No full textTerrorist risk in historical outdoor Open Areas (OAs), such as urban squares, is one of the most relevant conditions to explore for users’ safety, considering the attractiveness of these public outdoor spaces and their architectural, morphological and use-related features. Risk assessment tasks should be linked to risk mitigation analysis to support decision-makers in understanding which most probable terrorist threats and related effects could appear and how resilience-increasing strategies could be implemented while pursuing heritage preservation and ease of application. To this end, user factors in terms of behaviours in pre-emergency and evacuation conditions should be included, indeed. Nevertheless, structured frameworks to this end seem to be still lacking. This work proposes a novel, sustainable, user-centred framework for risk assessment and mitigation analysis to be applied in historical OAs such as squares. The framework relies on a simulation-based approach and incorporates tools validated within the BE S2ECURe project for scenario creation (preferring quick-but-reliable approaches) and evacuation process analysis, incorporating users’ behaviours. The framework is showcased on a relevant historical square, by defining and simulating pre and post-retrofit scenarios. In particular, mitigation strategies aimed at emergency management are mainly considered given their compatibility with preservation tasks. Results show how the framework can support local authorities in increasing resilience against terrorist threats
GENNEXT: The Next Generation of IR and Recommender Systems with Language Agents, Generative Models, and Conversational AI
No full textWe present GENNEXT, a workshop dedicated to exploring the integration of language agents, generative models, and conversational AI within information retrieval (IR) and recommender systems (RS). Building on the success of our recent RecSys’24 workshop, GENNEXT aims to advance discussions on the applications of language agents powered by Large Language Models (LLMs). The workshop will focus on enhancing interactivity between users and systems through multi-turn dialogues, improving creative content generation, advancing personalization, and enabling multifaceted, context-aware decision-making. For example, a language agent could respond to a query like “Suggest an eco-friendly food tour for a weekend in my city” by using a recommendation API to identify eateries specializing in sustainable or organic cuisine and a pollution API to ensure the selected routes have low air pollution levels. GENNEXT will bring together leading researchers and practitioners through keynotes, paper presentations, and a panel discussion. We invite full papers, short papers, and extended abstracts covering theoretical advancements, practical applications, and evaluation strategies for generative technologies in IR and RS. The workshop will address key themes such as conversational adaptation, generative content creation, and agentic tool usage, while tackling challenges like bias, data privacy, and hallucination risks. Overall, our main ambition is to foster dialogue on creating ethical, sustainable, and innovative systems while addressing emerging opportunities and risks in modern IR and RS
Hierarchical materials as tissue-like phantoms for photoacoustic imaging
No full textWe introduce a hierarchical manufacturing method for creating anatomical phantoms using water-in-elastomer micro-emulsions. This approach addresses the cost, ethical, and technical limitations of animal models, providing a reliable alternative for multimodal imaging and artificial intelligence development, particularly in photoacoustic imaging
