Documents published in Scipedia

• A Multigrid preconditioner for Discontinuous Galerkin Methods Applied to Numerical Weather Prediction

  P. Birken, A. Dedner, R. Klöfkorn, J. Kasimir

  ECCOMAS 2024.

  
  

  Abstract

  High fidelity fluid simulations have important applications in science and engineer ing, examples include numerical weather prediction and simulation aided design. Discontinuous Galerkin (DG) methods are promising high order discretizations for simulating unsteady com pressible fluid flow in three dimensions. Systems arising from such discretizations are often stiff and require implicit time integration. This motivates the study of fast, parallel, low-memory solvers for the resulting algebraic equation systems. For (low order) finite volume (FV) discretizations, multigrid (MG) methods have been suc cessfully applied to steady and unsteady fluid flows. But for high order DG methods applied to f low problems, such solvers are currently lacking. The lack of efficient solvers suitable for contemporary computer architectures inhibits wider adoption of DG methods. This motivates our research to construct a Jacobian-free precondi tioner for high order DG discretizations. The preconditioner is based on a multigrid method constructed for a low order finite volume discretization defined on a subgrid of the DG mesh. Numerical experiments on atmospheric flow problems show the benefit of this approach.

  Abstract
  High fidelity fluid simulations have important applications in science and engineer ing, examples include numerical weather prediction and simulation aided design. Discontinuous [...]

  • 0
  •  read

• Finite Element Simulation of a Rainfall Induced Shallow Landslide in an Experimental Hillslope with a Multiphase Porous Media Model

  M. Lazari, M. Camporese, L. Sanavia

  ECCOMAS 2024.

  
  

  Abstract

  This study presents the results of a FEM numerical simulation of a large scale physical model of a slope subjected to rainfall infiltration. The slope failure is modelled as a coupled variably saturated thermo-hydro-mechanical problem, using the Pastor-Zienkiewicz generalised plasticity model to obtain the soil’s mechanical response. Small strain and quasi static loading conditions are assumed, and plane strain conditions are adopted in the slope stability analysis. The hydraulic and mechanical parameters are calibrated based on the available experimental data. The numerical results are compared with the experimental data of the mechanical and the hydraulic responses up to failure.

  Abstract
  This study presents the results of a FEM numerical simulation of a large scale physical model of a slope subjected to rainfall infiltration. The slope failure is modelled [...]

  • 0
  •  read

• MPM analyses of impacts of dry and saturated granular masses against rigid obstacles

  M. Zerbi, P. Marveggio, C. Prisco

  ECCOMAS 2024.

  
  

  Abstract

  In recent years, the impact of saturated granular flows against rigid obstacles has been studied by using different numerical approaches. The very low compressibility of water causes numerical instabilities when impact problems are simulated. In this work, a sensitivity analysis has been done by using a Material Point Method code to assess the influence of fluid compressibility and front inclination on numerical results. When the mass front is inclined, fluid bulk modulus does not significantly affect the solution and can be reduced to speed up the computations and reduce spurious numerical oscillations.

  Abstract
  In recent years, the impact of saturated granular flows against rigid obstacles has been studied by using different numerical approaches. The very low compressibility of water [...]

  • 0
  •  read

• Studies on Effect of Interaction of Coaxial Jets on Thrust of Air-breathing Rocket

  T. Sawada, A. Oyama, Y. Maru, H. Manako

  ECCOMAS 2024.

  
  

  Abstract

  It is anticipated that the rocket-based combined cycle engine, which incorporates a rocket engine and an airbreathing engine, will offer enhanced propulsive capabilities compared to conventional rocket engines. The exhaust nozzle is a coaxial nozzle comprising a convergent divergent nozzle in the center and a convergent nozzle around it. This represents an unprecedented configuration for a rocket engine nozzle. The objective of this study is to numerically analyze the flow field near the nozzle exit and to elucidate the impact of jet interference on thrust to facilitate the detailed design of rockets. In this study, an airbreathing sounding rocket, currently under research and development at JAXA, is employed as the analysis target. The resulting calculation yielded the flow field data around the nozzle. When the central jet is over-expanded, the velocity and pressure distributions at the nozzle outlet undergo alterations due to the mutual effect of one jet pulling in the other jet. The combined thrust of the two nozzles activated simultaneously was found to be lower than the sum of the individual thrusts of the two cases in which only one of the nozzles was activated. Conversely, the thrust remains constant when the central jet is under-expanded.

  Abstract
  It is anticipated that the rocket-based combined cycle engine, which incorporates a rocket engine and an airbreathing engine, will offer enhanced propulsive capabilities compared [...]

  • 0
  •  read

• Data-driven Permeability Prediction of 3D Fibrous Microstructures

  D. Natarajan, T. Schmidt, S. Cassola, M. Nuske, M. Duhovic, D. May, A. Dengel

  ECCOMAS 2024.

  
  

  Abstract

  For the manufacturing process simulation of fiber-reinforced polymer composites, f low simulations have to be performed at multiple spatial scales which govern the flow through the fiber structures. Repetitive multiscale flow simulations are computationally expensive and time-consuming. In order to speed up the multiscale simulation workflow, fast machine learning surrogate models or emulators could be used to replace one or more of the flow simulations. In this work, feature-based emulators and geometry-based emulators are developed using neural networks for predicting the permeability of 3D fibrous microstructures based on a reference dataset (doi:10.5281/zenodo.10047095). The best model achieved a mean relative error of 8.33% on the test set with a significantly faster inference time compared to a conventional simulator.

  Abstract
  For the manufacturing process simulation of fiber-reinforced polymer composites, f low simulations have to be performed at multiple spatial scales which govern the flow through [...]

  • 0
  •  read

• Machine learning model for correlating microstructural features and macroscopic properties of heterogeneous composites

  H. Zhao, C. Shen, K. Wang, R. Mu, A. Wang

  ECCOMAS 2024.

  
  

  Abstract

  This study explores the use of machine learning (ML) models in predicting the macroscopic properties of heterogeneous composites. Traditional micromechanics parameters have limitations, thus ML models with and without feature engineering are utilized. For artificial neural network (ANN) models with feature engineering, microstructural descriptors from SEM images of nickel-based superalloys are used to predict hardness. 10 descriptors are selected to reduce the computational cost of the deep neural network (DNN) with the support of the shallow neural network (SNN), and accuracy is enhanced by incorporating two additional descriptors. The result surpasses existing physics-based models. Models without feature engineering employ a convolutional neural network (CNN) to predict the effective thermal conductivity of thermal insulation composite materials. The CNN model demonstrates accurate predictions for novel microstructures. ML models can achieve more efficient predictions than traditional methods, indicating their potential in advancing materials science. In summary, harnessing artificial intelligence to capture the scattering characteristics of heterogeneous materials enables both DNN and CNN models to achieve more efficient predictions compared to traditional methods. This highlights the potential of machine learning in advancing materials science and expediting the development of materials with desired properties.

  Abstract
  This study explores the use of machine learning (ML) models in predicting the macroscopic properties of heterogeneous composites. Traditional micromechanics parameters have [...]

  • 0
  •  read

• Quasi Gasdynamic Heterogeneous Model for Modeling a Mixture of Compressible Fluids

  I. Khaytaliev, E. Shilnikov, T. Elizarova

  ECCOMAS 2024.

  
  

  Abstract

  Quasi-gasdynamic type regularization is presented for a heterogeneous model of a two-fluid mixture of compressible fluids. This model allows to describe the flows of stiffened gases. The reduced four-equation model for dynamics of the heterogeneous compressible two fluid mixture with equations of state of a stiffened gas is considered. A further reduced form of this model with the excluded volume concentrations and a quadratic equation for the common pressure of the components can be called a quasi-homogeneous form. A finite difference algorithm is used, built with the finite volume method. By solving one and two dimensional test problems it is shown that the presented algorithm is a stable and reliable way to model fluid mixtures with strong shock waves.

  Abstract
  Quasi-gasdynamic type regularization is presented for a heterogeneous model of a two-fluid mixture of compressible fluids. This model allows to describe the flows of stiffened [...]

  • 0
  •  read

• Deep Reinforcement Learning implementation with Physics-Informed Neural Network for Heat Conduction Control

  N. Gonçalves, J. Rodrigues

  ECCOMAS 2024.

  
  

  Abstract

  As modern systems become more complex, their control strategy can no longer solely rely on measurement data gathered by instrumentation. Instead, it must also incorporate information derived from mathematical models. The complexity of system models can result in excessively long computation times, making the control process impractical. As a solution, surrogate models are implemented to provide estimates within an acceptable timeframe for decision-making purposes. The surrogate model can be a Physics-Informed Neural Network that is used to obtain the system state on the next time step; such information can be used with a Deep Reinforcement Learning algorithm to train a control strategy based on simulations, replacing the need for running direct numerical simulations. On this work, we explore a Deep Q-Learning strategy on 1D heat conduction problem in which temperature distribution feeds a control system to activate a heat source, aiming to obtain a constant, previously defined temperature value. The main goal is to stabilize the bar temperature at the middle point of it without recurring to numerical simulations.

  Abstract
  As modern systems become more complex, their control strategy can no longer solely rely on measurement data gathered by instrumentation. Instead, it must also incorporate [...]

  • 0
  •  read

• A second-order reproducing kernel for SPH method to enhance accuracy and consistency

  I. Wiragunarsa, L. Zuhal, T. Dirgantara, I. Putra

  ECCOMAS 2024.

  
  

  Abstract

  The application of smoothed particle hydrodynamics (SPH) encounters challenges related to consistency, stability, and accuracy. Inconsistencies in SPH arise from non-uniform particle distribution and a lack of neighboring particles at the boundary, leading to numerical instability and inaccurate particle approximations. Various methods have been proposed to address these issues. One such framework is the corrected SPH, designed to ensure consistency of the method. In this work, performance of some correction procedures are analysed through gradient calculations of a function. The root mean square error of the gradient approximation is analysed to justify the method’s convergence and accuracy

  Abstract
  The application of smoothed particle hydrodynamics (SPH) encounters challenges related to consistency, stability, and accuracy. Inconsistencies in SPH arise from non-uniform [...]

  • 0
  •  read

• A 3D Particle Model For Reinforced Concrete Fracture Aanalysis

  N. Azevedo, M. Farinha, S. Oliveira

  ECCOMAS 2024.

  
  

  Abstract

  A numerical model for the analysis of reinforced concrete structures must incorporate tools capable of representing the formation and propagation of cracks, their effect on the overall behavior of the structure, and the interaction between reinforcement and concrete. Detailed rigid particle models (PM) that take directly into consideration the physical mechanisms and the influence of the material aggregate structure have gained relevance and have shown to be able to predict, evaluate and understand cracking phenomena in concrete. The 3D particle models correlate well with experimental results from concrete specimens, particularly in terms of elastic response, peak values, fracture process and fracture location. This paper presents the 3D explicit formulation of steel reinforcement bars using discrete elements with cylindrical geometry. The incorporation of steel elements allows the particle model to be applied to the analysis of fracture in reinforced concrete structures. The rigid elements of cylindrical geometry interact with the concrete, modeled by spherical particles, through a contact interface. The model is validated in three-point beam bending tests, without transverse steel reinforcement. The numerical results obtained show that the proposed model correctly simulates the actual behavior, representing the fracture evolution process and the load displacement relationship for different steel ratios.

  Abstract
  A numerical model for the analysis of reinforced concrete structures must incorporate tools capable of representing the formation and propagation of cracks, their effect on [...]

  • 0
  •  read