Publications HAL du labo/EPI LMEE

2024

Theses

titre
Modeling of coupled vibration systems with fluid-structure interaction
auteur
Runze Zhang
article
Fluid mechanics [physics.class-ph]. Université Paris-Saclay, 2024. English. ⟨NNT : 2024UPAST136⟩
resume
This thesis addresses two key aspects of the vibrational dynamics of structures with fluid-structure interaction (FSI): the control of vibrations in structures immersed in a fluid and the conversion of fluid-induced vibrations into electrical energy. This dissertation presents our contributions to these two domains. To this end, the study first proposes an unit cell-based finite element model to predict vibration bandgaps in periodic composite plates under FSI conditions. By introducing a novel fluid-added mass matrix integrating Bloch boundary conditions, the fluid's inertial effects are incorporated into the bandgap analysis, enabling designing for vibration control in periodic plates submerged in liquids, achieving passive control.Based on this foundation, the research further explores the potential for actively tuning the vibration bandgaps of periodic composite plates submerged in liquids, which integrate connected piezoelectric sensors and actuators with feedback control. Therefore, an unit cell-based vibration bandgap tuning model with inertial fluid is developed, which integrates Bloch boundary conditions for both fluid and piezoelectric coupled solid domains. Then, the study reveals that in liquid environments, the fluid-added mass effect significantly impacts the bandgap characteristics of thin-walled structures, reducing the effectiveness of control strategy. Increasing the structure's self inertia or optimizing the arrangement of piezoelectric patches can mitigate this effect.On the other hand, the rivers and oceans are in constant motion, containing substantial kinetic energy. When fluid flows over a structural surface, the induced structural vibrations can be viewed as a potential source of clean and renewable energy. By utilizing the direct piezoelectric effect of piezoelectric materials, the kinetic energy of the fluid can be converted into usable electrical energy, enabling fluid energy harvesting. However, in such energy harvesting systems, significant FSI and electro-mechanical coupling effects are often accompanied by complex nonlinear dynamic behavior. The presence of these coupling effects complicates numerical simulations in this field, making it challenging, especially when considering practical applications where a deep understanding of these nonlinear behaviors and their impact on system performance is essential. Therefore, this thesis develops a full-scale finite element model to capture the strong local FSI behavior of complex thin-walled piezoelectric fluid energy harvesters (PFEH) involving microstructured transducers and non-uniform cantilevers, which are often ignored by simplified models. The research analyzes different energy harvester designs through numerical simulations, examining the influence of substrate cross-sectional shape, piezoelectric patch arrangement, and microstructure on the system's dynamic response and energy output efficiency.Finally, the study further enhance the power output of PFEHs using synergistic vortex generators composed of upstream double plates and downstream cylinder with a small spacing in dynamic water environments. With the synergistic effects of multi vortex generators, it is possible to achieve higher frequency and stable larger amplitude vibrations for the piezoelectric flag, thereby obtaining higher energy harvesting efficiency. Overall, the multi-physics coupling modeling for different FSI conditions proposed in this study not only effectively predict and control structural vibrations in fluid environments but also provide a theoretical foundation and technical support for the development of efficient piezoelectric energy harvesting systems.
Accès au texte intégral et bibtex
https://theses.hal.science/tel-04875847/file/2024UPAST136.pdf BibTex
titre
Reduction of non-linear thermal models by modal reduction method applied to the modeling of complex electronic components.
auteur
Fatme Mustapha
article
Physique mathématique [math-ph]. Université Paris-Saclay, 2024. Français. ⟨NNT : 2024UPAST095⟩
resume
The miniaturization of electronic components is accompanied by an augmentation in the density of dissipated thermal power : Thermal science has become a limiting criterion during the design of electronic boards. The emergence of new and more complex technologies requires a re-evaluation of traditional methods of electronics thermal modeling. As such, Thales group team and Thermal and Energy team of the Laboratory of Mechanics and Energetics of Évry have been collaborating in order to evaluate a modal reduction method. The modal methods consist of finding the solution in the form of a weighted sum of elementary fields called modes. These modes are calculated by numerically solving a problem with the eigenvalues. When a small number of modes is enough to reconstruct correctly the temperature field, then these methods reduce effectively the problem posed and the associated computation times. However, most reduction methods focus on linear models, whereas the microchip, which forms the active and essential part of the electronic component, is made of materials whose conductivity decreases significantly with temperature, leading to a notable increase in the maximum temperature. Incorporating this non-linearity requires recalculating the temperature and conductivity matrix in the physical space and then assessing its evolution in the modal space at each iteration. These operations significantly reduce the efficiency of the reduced model in terms of computational time. Therefore, the goal of this thesis is to propose a solution that efficiently addresses nonlinear problems using reduced models, while maintaining the geometric and functional complexity. This approach allows the simulation of different cooling configurations for the same component, considering multiple activation combinations that require numerous additional calculations. In parallel to this theoretical work, a software engineering work was carried out in order to ensure the technological transfer from the laboratory to the industry, in particular the connection between the software suite used by Thales and the research code developed by the LMEE.
Accès au texte intégral et bibtex
https://theses.hal.science/tel-04723372/file/2024UPAST095.pdf BibTex
titre
Use of machine learning for the construction of seismic fragility curves
auteur
Anh-Dung Tran
article
Génie civil. Université Paris-Saclay, 2024. Français. ⟨NNT : 2024UPAST017⟩
resume
The evaluation of seismic risks for structures is necessary to prevent human and material losses in the event of natural disasters. The construction of seismic fragility curves, the tool that defines the probability of failure of a structure based on earthquake intensity, plays a significant role in this assessment. Generally, the method of constructing these curves requires costly procedures, often involving finite element analyses, which require considerable time and computing resources. This difficulty hinders the real-time use of this tool or its application for more common structures.Machine learning has experienced a remarkable development with its application in various fields. It is recognized as a powerful tool for modeling complex relationships between inputs and outputs from data. An innovative approach is emerging: the use of machine learning models to predict structural responses to earthquakes. Therefore, the main motivation of the thesis is to study the application of machine learning to generate seismic responses for seismic risk assessment, specifically for constructing fragility curves.The study begins with a literature review, which presents seismic fragility curves and the challenge related to the computational burden for their construction. After a brief introduction to machine learning, the first chapter focuses on its application to model seismic responses of structures. Given the sometimes contradictory wide range of characteristics, it is necessary to propose practical procedures with the most relevant and easy-to-implement features.The thesis addresses this issue and proposes an innovative approach to efficiently use and select the acceleration response spectrum sampled at different periods to build machine learning models. Two procedures, named PRO-LIN and PRO-NONLIN, are proposed for linear and nonlinear structures respectively. To validate these proposals, tests are performed on linear and nonlinear structures combined with synthetic records, demonstrating a significant reduction in simulation time while maintaining accuracy in constructing fragility curves.Although the initial validations of the procedures are conducted with synthetic records, validation with real records is essential to validate these proposals. The records are selected according to the conditional spectrum from databases. Finally, another validation aims to test the proposed procedures with existing databases in the literature. The first database concerns a linear reinforced concrete structure. The second database concerns nonlinear moment-resistant steel frames.In conclusion, based on the results obtained in the studies, this work highlights the effectiveness of the proposed procedures in the thesis. These procedures effectively improve the construction of seismic fragility curves and seismic risk assessment.
Accès au texte intégral et bibtex
https://theses.hal.science/tel-04721028/file/2024UPAST017.pdf BibTex

2023

Theses

titre
Numerical modeling of turbulence and atmospheric dispersion in low wind-speed conditions and urban environments.
auteur
Boulos Alam
article
Mécanique des fluides [physics.class-ph]. Université Paris-Saclay, 2023. Français. ⟨NNT : 2023UPAST179⟩
resume
This thesis is situated in the context of atmospheric dispersion modeling, particularly in the presence of low winds. Atmospheric pollution sources, often located near the ground and influenced by complex obstacles, generate high concentrations of pollutants nearby, resulting in significant concentration fluctuations. Low winds, typically associated with stable atmospheric conditions, pose a specific challenge in modeling pollutant dispersion, requiring a thorough analysis of meteorological data and adaptation of prediction models. To address this complex challenge, the use of Computational Fluid Dynamics (CFD) is necessary, although further research is needed to validate its effectiveness in the near-field and in the presence of low winds. The Code_Saturne® software (EDF R&D) is selected due to its proven efficiency in simulating atmospheric pollutant dispersion. This thesis is divided into three distinct phases : the first phase focuses on the fundamentals of atmospheric dispersion, exploring the impact of various parameters such as the atmospheric boundary layer structure, atmospheric turbulence, and atmospheric stability. These elements play a crucial role in how pollutants disperse in the air. The second phase details the methodology used in Code_Saturne for conducting simulations, including the turbulence models employed and the criteria for evaluating these models. In addition to traditional isotropic models, this research investigates the use of anisotropic turbulence models to study dispersion in various contexts. The third phase of the thesis concentrates on the evaluation of different turbulence models and velocity-scalar correlations using observations conducted in urban environments under neutral and stable atmospheric conditions. Finally, the last phase of the research explores conditions of low and stable winds, typically characterized by wind speeds below 2 m/s and random wind variations. This phase examines the meandering patterns in pollutant dispersion and assesses the limitations of analytical and CFD models in predicting concentration in such conditions. To this end, a URANS model is developed and evaluated. Ultimately, a segmented Gaussian method is devised to compare the results with CFD predictions and field observations.
Accès au texte intégral et bibtex
https://theses.hal.science/tel-04721194/file/2023UPAST179.pdf BibTex

2022

Theses

titre
Evaluation of environmental impacts by a Dynamic Spatialized LCA
auteur
Patrice Megange
article
Mécanique [physics.med-ph]. Université Paris-Saclay, 2022. Français. ⟨NNT : 2022UPAST171⟩
resume
Life Cycle Assessment (LCA) is a holistic approach to environmental analysis that is widely used in the 21st century to evaluate the potential environmental impacts of products, systems and processes. Studies on the availability and reliability of the data used have shown that uncertainties generated during these environmental assessments reduce the robustness of the results and the analysis chain. To overcome this issue, recent studies recommend the introduction of spatio-temporal specificities where solutions such as Dynamic Life Cycle Assessment or Regionalized Life Cycle Assessment have demonstrated their relevance. However, these LCA methods have some limitations in terms of operationality and versatility.The aim of this thesis is to develop a reliable and conceptualized tool to optimize LCA for assessing local impacts such as human inhalation toxicity. The approach is based on the coupling of complementary tools integrating the dynamics of processes, the pollutant emission kinetics and determinant spatio-temporal specificities. This results in a new method called Site-dependent Dynamic Life Cycle Assessment (ACVDSd).The ACVDSd components are: (i) ESPA (Enhanced Structural Path Analysis), a Dynamic Life Cycle Inventory method to integrate the spatial and temporal disaggregation of economic and elemental flows with distribution algebra,; (ii) a Gaussian modeling of pollutants atmospheric dispersion to evaluate the concentrations of received toxic substances; (iii) USEtox, a consensus model for the evaluation of the human toxicity and ecotoxicity impact.In this research work a methodology was developed to ensure the genericity of the model. Applied to the impact of human toxicity, it aims to optimize the fate and absorption factors by integrating relevant spatial and temporal characteristics. As air pollution contributes to the contamination of environments such as soils and waters, this method will also have a beneficial impact on the possible evaluation of ecotoxicity in these same environments.In order to analyze the viability of the proposed methodology, a feasibility study is performed. It consists in the observation from results of a human toxicity impact assessment using the ACVDSd. The studied scenario is the manufacturing stage of double glazing windows in PVC frame. It addresses to a real estate program and the expectations of the Environmental Regulation 2020. In order to generalize the results obtained with a classical Gaussian dispersion model, specifically adapted for a flat terrain, a Gaussian puff transfer-diffusion model (CALPUFF) is used for a rough terrain configuration.The accuracy and the spatio-temporal specificities highlighted by this research work open several other interesting perspectives such as the spatial heterogeneity of populations or an ecotoxicity impact study. The use of a Lagrangian dispersion model can be considered for the evaluation of a global impact such as climate change.
Accès au texte intégral et bibtex
https://theses.hal.science/tel-03945145/file/2022UPAST171.pdf BibTex
titre
Structural health monitoring based on operational modal analysis
auteur
Duc Tuan Ta
article
Mécanique [physics]. Université Paris-Saclay, 2022. Français. ⟨NNT : 2022UPAST120⟩
resume
Structural health monitoring (SHM) is primordial for safe use and is essential for sustainable development. Among existing methods for SHM, vibration-based methods are the most commonly used. Operational modal analysis (OMA) is suitable for real structures as it offers several advantages: low cost, normal use of structures, and continuous monitoring. However, it has some main obstacles: (i) uncertainty in identified modal parameters due to unmeasured and uncontrolled operational excitations; (ii) underdetermined problems when the number of measured responses is less than that of active modes; (iii) the relationship between the damage in terms of change in mechanical properties like mass and stiffness and the change in modal parameters, is not straightforward, and it often goes through finite element update steps resulting in computational burden; (iv) in reality, there may be several damages in a structure, and the detection of multiple damages is not obvious. Therefore, the objectives of the thesis are: (i) overview of efficient and popular methods for operational modal analysis and damage identification; (ii) propose improvements to existing methods or a novel method that can deal with underdetermined cases; (iii) develop a procedure for rapid damage detection based on a simplified relationship between damage and changes in modal parameters; (iv) introduce an enhanced procedure for multiple damage detection in structures.To achieve these objectives, the obtained results of the thesis can be briefly summarized in the following four contributions.The first contribution is an improvement of the existing modal identification technique based on the PARAllel FACtor (PARAFAC) decomposition in time domain. The third-order tensor of the covariance of responses is first decomposed into components corresponding to structural modes or harmonic components. A minimum length of autocovariance functions using natural periods and damping factors is suggested to distinguish between harmonics and structural modes accurately.The second contribution is the development of a novel method for modal identification based on PARAFAC decomposition in frequency domain. Using the PARAFAC decomposition, a third-order tensor in frequency constructed from Power Spectral Density (PSD) of responses is first decomposed into rank-1 tensors that can be structural modes or harmonic components. The auto-PSD function of each rank-1 tensor is then used to identify modal parameters, while spectral kurtosis values are used for the distinction of structural modes and harmonics.The third contribution is devoted to the proposal of an efficient method for the rapid detection and quantification of a single local change in the mass and/or stiffness of like-beam structures using identified modal parameters. This contribution considers the relationship between local changes in the mass and/or stiffness of a beam and its natural frequency shift and mode shape, and explicitly gives an analytical expression. Based on the proposed expression, linear regression is applied to obtain accurate results of the change in the mass/stiffness of the beam.The fourth contribution aims to extent the previous damage identification procedure for multiple local changes in mass and/or stiffness. Comparison between natural frequency shifts obtained directly from the analytic expression established in the former contribution instead of using FEM and measured ones allows multiple damages to be identified using Bayesian inference. The proposed identification of damages becomes rapid because it skips the computational cost caused by FEM simulations.All the above contributions have been validated by numerical simulations and experimental laboratory tests.
Accès au texte intégral et bibtex
https://theses.hal.science/tel-03835377/file/2022UPAST120.pdf BibTex
titre
Computation homogenization of heterogeneous materials' thermal conductivity by the Finite Difference Method
auteur
Ziming Xiong
article
Solid mechanics [physics.class-ph]. Université Paris-Saclay, 2022. English. ⟨NNT : 2022UPAST045⟩
resume
Nowadays, heterogeneous materials are in-creasingly used for their superior overall properties, suchas porous media, which are widely used in the electronicsand biomedical industries, so determining the equivalentthermal conductivity (ETC) of heterogeneous materials isessential for the correct design of industrial equipment thatmay be subjected to severe thermal loads during use.The main objective of this thesis is to calculate the ho-mogenization of the thermal conductivity of heteroge-neous materials using the finite difference method. Voxelwas chosen for modeling heterogeneous materials and the Günter scheme will be employed as the primary tech-nique for anisotropic thermal diffusion problems. Thetwo-dimensional Günter system is re-demonstrated in thisthesis, along with an extension to the three-dimensionalmodel, as well as methods for loading periodic and mixeduniform boundary conditions. The three methods (FDM,FEM, and FEM+pixel(voxel)) are compared for 2D RVEssuch as crosses, circles, and ellipses and for 3D RVEs suchas spheres and cylinders. It is discovered that the devel-oped FDM produces results that are consistent with thoseof FEM and FEM+pixel(voxel) and that the FDM outper-forms FEM+pixel(voxel) in terms of convergence speed.This method has also been applied to sintered silver ma-terials for the study of equivalent thermal conductivity.Comparisons between the two methods (FDM and FEM)are carried out for the classical unit cells such as simple cu-bic, body-centered cubic, and face-centered cubic, as wellas the silver-based stochastic model. The developed finitedifference algorithm is valid, and consistent results are ob-tained. In addition to the Günter scheme, a 5-point modeland an integral model have also been developed inspiredby the Günter scheme.For high-performance computing, the Eigen library andthe Pardiso library are also detailed in the thesis. Both li-braries contain both direct and iterative solutions for solv-ing linear equations. However, while Eigen allows for parallel computation of only the iterative solution, Pardisoallows for parallel computation of both approaches, andthe parallelism is significantly superior than that of Eigen.While Eigen is more straightforward to construct and morepowerful, Pardiso is faster at tackling complex problems.
Accès au texte intégral et bibtex
https://theses.hal.science/tel-03675234/file/2022UPAST045.pdf BibTex
titre
Modelling of adhesive frictional contact problems for soft matters
auteur
Libang Hu
article
Mechanics of materials [physics.class-ph]. Université Paris-Saclay, 2022. English. ⟨NNT : 2022UPAST032⟩
resume
Dynamic frictional contact and recoverable interfacial adhesion involving soft materials represent a frequent contact phenomenon. In the numerical modelling field, constructing of a model capable of addressing contact with friction and recoverable adhesion has always been a challenging topic. In this thesis, an extended, ready-to-implement 3D model for quasi-industrial problems of contact with friction and recoverable interface adhesion between soft material is formulated using the Raous-Cangémi-Cocou (RCC) interface model and a bi-potential based resolution method. The RCC model proposes a straightforward description of the interface adhesion based on a local scalar parameter, both normal and tangential effects are taken into account by the adhesive interface model, involving both the process of bonding and de-bonding of the interface links. This adhesive model has been implemented within the bi-potential method, based on a set of extended unilateral and tangential contact laws. We combine the 3D extended adhesive interface model with different hyperelastic models to investigate large deformation contact problems under various adhesive interface conditions. Such as, Blatz-Ko material model for large deformation contact problems under isotropic and orthotropic adhesive interface conditions; The mechanical behaviour of biological soft tissues with surface adhesion is investigated by using the Holzapfel-Gasser-Ogden (HGO)+Yeoh anisotropic hyperelasticity model. To illustrate the capability of the implemented model, we set up various test cases in each chapter to explore adhesive contact in normal, tangential and mixed directional scenarios for different material models and interface conditions, which brings us closer to quasi-industrial modelling situations.
Accès au texte intégral et bibtex
https://theses.hal.science/tel-03630458/file/2022UPAST032.pdf BibTex

2020

Theses

titre
Contribution of reduced models for the thermal characterization of building materials : in situ measurements of insulators, and multi-scale study of dry wood
auteur
Ana Chavez Castillo
article
Matériaux composites et construction. Université Paris-Saclay, 2020. Français. ⟨NNT : 2020UPAST040⟩
resume
In numerical thermal simulation, the inverse problem consists in finding one or more parameters of the discretized heat equation from temperature measurements. This is a complex procedure that often remains limited to simple geometry. The idea is then to use modal-type thermal reduced models, which will considerably reduce the number of unknowns while maintaining satisfactory accuracy over the entire modelled domain. These models will then allow to extend the technique of inverse problems to any type of geometry, whatever its complexity.The objective of this thesis work is to evaluate the efficiency of such a method for an application related to building thermics, in which one seeks to identify the properties of insulating materials (thermal capacity and conductivity).The main work has been the application of this technique for an in situ measurement, using a hot wire probe, which has so far been unsuitable for thermal insulation.A second application of this technique to solve inverse problems by reduced model is the characterization of a bio-sourced material from tomographic surveys at the microscopic scale.For these two applications, the digital developments carried out have allowed the realization of encouraging first experimental trials.
Accès au texte intégral et bibtex
https://theses.hal.science/tel-03184785/file/2020UPAST040.pdf BibTex
titre
Numerical Simulations of the shock wave-boundary layer interaction in complex geometries
auteur
Adarsh Gupta
article
Other [cond-mat.other]. Université Paris-Saclay, 2020. English. ⟨NNT : 2020UPAST013⟩
resume
The objective of the present thesis work is to provide a better insight of the SWBLI unsteadiness due to the low-frequency streamwise oscillations of the separation bubble. To investigate this low frequency motion, DNS of the interaction between the shock wave and laminar boundary layer in complex geometries has been carried out. To perform those simulations, a modified numerical approach for curvilinear coordinate, implemented in an in-house parallel (MPI) Finite-Volume based DNS/LES solver (CHORUS) developed at LIMSI-CNRS is used.The first part of the thesis is the validation of the modified numerical approach. The influence of the mesh distortion has been analyzed from several test cases. The errors introduced by different types of deformation for the three test cases dealing with advection, turbulence, and shock wave were identifiable. The errors created by deformation of the mesh are found comparatively low if the control volumes stay close to a parallelepiped. In some cases, a significant rise has been seen due to the introduction of the non-orthogonality of the mesh.The second part is the validation of code in the framework of supersonic flows around a compression corner which is the core of the present dissertation. The validation studies have been carried out for the case of both inviscid and viscous flows over a compression ramp and the comparison with theoretical as well as numerical data has been presented. This comparison has shown that the results obtained with CHORUS code are in good agreement with the reference data. However, those studies are rather old and a lot of progress has been made in numerical methods for high-speed flow simulations. Unfortunately, there are only a few recent studies concerning simulations or experiments of fully laminar flow around ramps or other complex geometries that could have helped to assess Chorus’ ability to compute such flows. It has then been decided to create our own test case using an extensively tested supersonic flow solver, rhoCentralFoam of the OpenFOAM open-source numerical package. The results obtained provided the difference in the two numerical approaches and allowed us to consider Chorus as validated for DNS of compressible flows with shocks in complex geometries.Consequently, the last chapter deals with the physical analysis of the flow created by a laminar boundary layer developing around two geometries: a classical compression ramp and a compression-expansion ramp. As said earlier, the goal of those simulations was to determine whether the low-frequency oscillations of the recirculation zone can be related to the coherent structures in the incoming boundary layer. The results have demonstrated that, for both configurations, the separation shock IS NOT subjected to longitudinal oscillations. However, when analysing the spectra from probes in the vicinity of the separation point, it has appeared that all the frequency information is contained in those temporal signals. The conclusion of this study is that the absence of oscillations in the laminar case is not, as originally thought, due to the absence of coherent structures in the incoming boundary layer but rather to the fact that, in the laminar case, the separation bubble extent is too large. As a consequence, even if the perturbations that make the bubble oscillate in the turbulent case are present for laminar boundary layer, they are damped in such a way that they are not able to move the shock system and/or the recirculation zone. The next step to this study would be to reduce either the freestream Mach number or the ramp angle in order to have a smaller recirculation bubble and check if the motion appear in that case.
Accès au texte intégral et bibtex
https://theses.hal.science/tel-03152378/file/2020UPAST013.pdf BibTex

2018

Theses

titre
Reducing a thermal problem by the modal sub-structuring method. Application to the modeling of complex electronic packages
auteur
Sébastien Grosjean
article
Sciences de l'ingénieur [physics]. Université Paris - Saclay, 2018. Français. ⟨NNT : 2018SACLE039⟩
resume
An electronic component heats. Thermal optimization of the design of electronic packages and of their activation and de-activation duration is essential to cap the temperatures reached. Thus we have to be able to finely and quickly predict the thermal evolution of a component set in its surroundings for multiples scenarios of use. Classical simulations, like finite elements ones, ate too costly in terms of computing time for complex packages, we have to reduce the size of the model. Modal methods consist in seek the solution as a weighted sum of elementary fields, labelled modes. These modes are computed by solving eigenvalue problems. But these methods don't work for packaging containing dozens of components on a printed circuit board.The modal sub-structuring method is an extent of classical modal methods used to overcome their limitations. The principle is to disassemble the system into elementary entities (the sub-structures), to compute the modes of each of these entities and to reassemble them to solve the original problem. Preliminary work has shown the relevance of this method on the first level (electric component). The thesis is aimed to extend these principles to superior levels. The industrial purpose of this work is to simulate, thanks to a reduced model, the thermal behavior of an electronic card.Beside the industrial aspect, this work raises fundamental issues like the definition of the junction conditions between the sub-structures in the modal space and the adaptation of the reducing technique of the sub-structures in view of their pairing.The thesis will give answers to these questions using an electronic card as a study support.This work could pave the way of the industrial for the modeling of multi-modules systems.
Accès au bibtex
BibTex

2017

Theses

titre
Sensors networks optimization for the characterization of atmospheric releases source
auteur
Hamza Kouichi
article
Mécanique des fluides [physics.class-ph]. Université Paris Saclay, 2017. Français. ⟨NNT : ⟩
resume
The main objective of this study is to define the methods required to optimize a monitoring network designed for atmospheric source characterization. The optimization consists in determining the optimal number and locations of sensors to be deployed in order to respond to such needs. In this context, the optimization is performed for the first time by a coupling between the data inversion technique named "renormalization" and the metaheuristic optimization algorithms. At first, the inversion method was evaluated for a point source, and then have allowed to define optimality criteria for networks design. In this study, the optimization process was evaluated in experiments carried out in flat terrain without obstacles (DYCE) and in an idealized urban environment (MUST). Three problems were defined and tested based on these experiments. These problems concern (i) the determination of the optimal network size for source characterization, for which a cost function (standard errors) estimating the gap between observations and modeled data, has been minimized; (ii) the optimal design of a network to retrieve an unknown point source for a particular meteorological condition. In this context, an entropy cost function has been maximized in order to increase the information’s amount provided by the network; (iii) the determination of an optimal network to reconstruct an unknown point source for multiple meteorological configurations. For this purpose, a generalized entropic cost function that we have defined, has been maximized. For these all problems, optimization is ensured within the framework of a combinatorial optimization approach. The determination of the optimal network size (problem 1) was highly sensitive to experimental conditions (source height and intensity, stability conditions, wind speed and direction, etc.). We have noted that the networks performance is better for a dispersion on flat terrain compared to the urban environments. We have also shown that different networks architectures can converge towards the same optimum (approximate or global). For unknown sources reconstruction (problems 2 and 3), the entropic cost functions have proven to be robust and allowed to obtain optimal networks (for reasonable sizes) capable of characterizing different sources for one or multiple meteorological conditions.
Accès au texte intégral et bibtex
https://hal.science/tel-01593834/file/kouichi_2017.pdf BibTex

2006

Theses

titre
Développement d'algorithmes et d'un code de calcul pour l'étude des problèmes de l'impact et du choc
auteur
Benoit Magnain
article
Mécanique [physics.med-ph]. Université d'Evry-Val d'Essonne, 2006. Français. ⟨NNT : ⟩
resume
In many problems of solid mechanics, taking into account the frictional contact plays an important role. Good prediction of the results of contact with friction becomes then a prevailing element for the industry. However, few problems can be solved analytically. It is then necessary to develop numerical methods adapted to such issues. In this work, an extension of the bi-potential method, proposed by de Saxcé & Feng, is presented for the modeling of impact problems involving several deformable bodies within large deformations framework.To this purpose, we choose a first order time integrator instead of classical second order integrators (Newmark, HHT, . . .). This choice avoids to take account for the acceleration which is undefined at the impact instant. The model so developed combines the bi-potential method to solve contact problems and a first order scheme for the time integration. This work leads to develop the finite element code FER/Impact. Different numerical applications illustrate clearly the validity and efficiency of the method. A special attention is paid to the quantification of dissipated energy by friction.
Accès au texte intégral et bibtex
https://theses.hal.science/tel-00369905/file/these_magnain.pdf BibTex

2002

Theses

titre
Numerical simulation of three-dimensional flows in a rotating lid-cylinder cavity
auteur
Emerson Barbosa
article
Dynamique des Fluides [physics.flu-dyn]. Université d'Evry-Val d'Essonne, 2002. Français. ⟨NNT : ⟩
resume
The flow in a cylindrical tank filled with an incompressible viscous fluid is considered. The motion is due to the rotation of one lid at constant angular velocity. Recent three-dimensional computations have been recently performed for aspect ratios h>1,75. The aim of this work is the numerical investigation of the symmetry-breaking of the flow in a cavity of reduced aspect ratio (h=1 and 1,5) and Reynolds numbers Re<8500. The resolution of the Navier-Stokes equations in velocity-pressure formulation written in cylindrical coordinates is accomplished using second-order finite differences methods. Two methodologies have been applied to study the loss of axisymmetry. The governing equations cast in perturbation form with respect to an axisymmetric steady base flow, in linearized form, are solved starting from a randomly distributed initial perturbation. This linear stability analysis allows for the determination of the first threshold and highlights the most unstable azimuthal mode kc. The integration of the full Navier-Stokes equations is carried out starting from initial conditions that may be either a randomly distributed initial perturbation (i) or a rest state (ii) or an intantaneous field from a previous computation (iii). The nonlinear stability analysis (i) shows that a supercritical Hopf bifurcation involving a dominant azimuthal mode kc occurs when Re exceeds a first threshold. Solutions of the first branch are periodic, having associated period T, and their spatial structure depends on the value kc. Axisymmetry is broken beyond the bifurcation only when kc is not zero. In this case, the result is a wave which rotates around the axis with a period TRWkc=kc*T. The flow undergoes then subsequent bifurcations as Re is increased. The characteristics of the bifurcated rotating waves are presented in details. Starting from other initial conditions (ii)--(iii) highlights regions of hysteresis: the reached branch depends on the way the initial kinetic energy is distributed.
Accès au texte intégral et bibtex
https://theses.hal.science/tel-00354646/file/cuve.pdf BibTex

2000

Theses

titre
Atmospheric turbulent flows modelling in order to study aerodynamic effects induced by the wind on civil engineering structures
auteur
Grégory Turbelin
article
Mécanique des fluides [physics.class-ph]. Université d'Evry Val d'Essonne, 2000. Français. ⟨NNT : ⟩
resume
The aim of this thesis is to compute atmospheric turbulent flows in order to study aerodynamic effects induced by the wind on civil engineering structures. A two-dimensional analysis has been carried out to provide a description of the phenomena involved when the wind acts on a bluff body elongated in the across wind direction. In the discussion, the wind has been assumed as an incompressible turbulent flow governed, in the atmospheric boundary layer, by the Navier-Stokes equations. The turbulent stresses have been determined by a first order turbulence model. The key parameters that have been taken into accounts are the surface roughness, the topogaphy, the obstructions and/or the development of large-scale organised vortices. The computational procedure has been performed with a finite-element software, CASTEM 2000, in which some uncommon methodology, like wall-functions based upon roughness length concept, have been incorporated. Validation tests have been performed for well-known pratical cases. The results show that the standard k-epsilon turbulence model is unable to simulate unsteady flows over changing terrains, with organised vortices. The more sophisticated RNG k-epsilon model appears to be more adapted to this kind of complex flows. Therefore, the latter model has been used to improve analytic formulations of buffeting forces. The existing quasi-steady formulation, commonly used in buffeting analysis, shows its limitations in the case of bluff bodies gust entry. A new approach is suggested which includes, through modified admittance functions, the turbulence induced by the structure itself. It has been found that the shape of these functions depends on both the creation of separation bubbles and the formation of vortices in the wake of the body.
Accès au texte intégral et bibtex
https://univ-evry.hal.science/tel-02359884/file/turthese.pdf BibTex