Coarse grid projection (CGP) methodology is a novel multigrid method for systems involving decoupled nonlinear evolution equations and linear elliptic Poisson equations. The nonlinear equations are solved on a fine grid and the linear equations are solved on a corresponding coarsened grid. Mapping operators execute data transfer between the grids. The CGP framework is constructed upon spatial and temporal discretization schemes. This framework has been established for finite volume/difference discretizations as well as explicit time integration methods. In this article we present for the first time a version of CGP for finite element discretizations, which uses a semi-implicit time integration scheme. The mapping functions correspond to the finite-element shape functions. With the novel data structure introduced, the mapping computational cost becomes insignificant. We apply CGP to pressure correction schemes used for the incompressible Navier Stokes flow computations. This version is validated on standard test cases with realistic boundary conditions using unstructured triangular meshes. We also pioneer investigations of the effects of CGP on the accuracy of the pressure field. It is found that although CGP reduces the pressure field accuracy, it preserves the accuracy of the pressure gradient and thus the velocity field, while achieving speedup factors ranging from approximately 2 to 30. Exploring the influence of boundary conditions on CGP, the minimum speedup occurs for velocity Dirichlet boundary conditions, while the maximum speedup occurs for open boundary conditions. We discuss the CGP method as a guide for partial mesh refinement of incompressible flow computations and show its application for simulations of flow over a backward facing step and flow past a cylinder. / Master of Science / Coarse Grid Projection (CGP) methodology is a new multigrid technique applicable to pressure projection methods for solving the incompressible Navier-Stokes equations. In the CGP approach, the nonlinear momentum equation is evolved on a fine grid, and the linear pressure Poisson equation is solved on a corresponding coarsened grid. Mapping operators transfer the data between the grids. Hence, one can save a considerable amount of CPU time due to reducing the resolution of the pressure filed while maintaining excellent to reasonable accuracy, depending on the level of coarsening.

Least Square Kinetic Upwind Method (LSKUM) belongs to the class of mesh-less method that solves compressible Euler equations of gas dynamics. LSKUM is kinetic theory based upwind scheme that operates on any cloud of points. Euler equations are derived from Boltzmann equation (of kinetic theory of gases) after taking suitable moments. The basic update scheme is formulated at Boltzmann level and mapped to Euler level by suitable moments. Mesh-less solvers need only cloud of points to solve the governing equations. For a complex configuration, with such a solver, one can generate a separate cloud of points around each component, which adequately resolves the geometric features, and then combine all the individual clouds to get one set of points on which the solver directly operates. An obvious advantage of this approach is that any incremental changes in geometry will require only regeneration of the small cloud of points where changes have occurred. Additionally blanking and de-blanking strategy along with overlay point cloud can be adapted in some applications like store separation to avoid regeneration of points. Naturally, the mesh-less solvers have advantage in tackling complex geometries and moving components over solvers that need grids. Conventionally, higher order accuracy for space derivative term is achieved by two step defect correction formula which is computationally expensive. The present solver uses low dissipation single step modified CIR (MCIR) scheme which is similar to first order LSKUM formulation and provides spatial accuracy closer to second order. The maximum time step taken to march solution in time is limited by stability criteria in case of explicit time integration procedure. Because of this, explicit scheme takes a large number of iterations to achieve convergence. The popular explicit time integration schemes like four stages Runge-Kutta (RK4) are slow in convergence due to this reason. The above problem can be overcome by using the implicit time integration procedure. The implicit schemes are unconditionally stable i.e. very large time steps can be used to accelerate the convergence. Also it offers superior robustness. The implicit Lower-Upper Symmetric Gauss-Seidel (LU-SGS) scheme is very attractive due to its low numerical complexity, moderate memory requirement and unconditional stability for linear wave equation. Also this scheme is more efficient than explicit counterparts and can be implemented easily on parallel computers. It is based on the factorization of the implicit operator into three parts namely lower triangular matrix, upper triangular matrix and diagonal terms. The use of LU-SGS results in a matrix free implicit framework which is very economical as against other expensive procedures which necessarily involve matrix inversion. With implementation of the implicit LU-SGS scheme larger time steps can be used which in turn will reduce the computational time substantially. LU-SGS has been used widely for many Finite Volume Method based solvers. The split flux Jacobian formulation as proposed by Jameson is most widely used to make implicit procedure diagonally dominant. But this procedure when applied to mesh-less solvers leads to block diagonal matrix which again requires expensive inversion. In the present work LU-SGS procedure is adopted for mesh-less approach to retain diagonal dominancy and implemented in 2-D and 3-D solvers in matrix free framework. In order to assess the efficacy of the implicit procedure, both explicit and implicit 2-D solvers are tested on NACA 0012 airfoil for various flow conditions in subsonic and transonic regime. To study the performance of the solvers on different point distributions two types of the cloud of points, one unstructured distribution (4074 points) and another structured distribution (9600 points) have been used. The computed 2-D results are validated against NASA experimental data and AGARD test case. The density residual and lift coefficient convergence history is presented in detail. The maximum speed up obtained by use of implicit procedure as compared to explicit one is close to 6 and 14 for unstructured and structured point distributions respectively. The transonic flow over ONERA M6 wing is a classic test case for CFD validation because of simple geometry and complex flow. It has sweep angle of 30° and 15.6° at leading edge and trailing edge respectively. The taper ratio and aspect ratio of the wing are 0.562 and 3.8 respectively. At M∞=0.84 and α=3.06° lambda shock appear on the upper surface of the wing. 3¬D explicit and implicit solvers are tested on ONERA M6 wing. The computed pressure coefficients are compared with experiments at section of 20%, 44%, 65%, 80%, 90% and 95% of span length. The computed results are found to match very well with experiments. The speed up obtained from implicit procedure is over 7 for ONERA M6 wing. The determination of the aerodynamic characteristics of a wing with the control surface deflection is one of the most important and challenging task in aircraft design and development. Many military aircraft use some form of the delta wing. To demonstrate the effectiveness of 3-D solver in handling control surfaces and small gaps, implicit 3-D code is used to compute flow past clipped delta wing with aileron deflection of 6° at M∞ = 0.9 and α = 1° and 3°. The leading edge backward sweep is 50.4°. The aileron is hinged from 56.5% semi-span to 82.9% of semi-span and at 80% of the local chord from leading edge. The computed results are validated with NASA experiments

L'objectif principal de cette thèse concerne la mise en oeuvre d'une méthoded'éléments finis stabilisés pour la simulation des plasmas de fusion. Pour cela,nous avons d'abord dérivé les modèles magnétohydrodynamiques depuis lemodèle cinétique. Les modèles MHD sont généralement utilisés pour simuler lesinstabilités macroscopiques des plasmas. Nous nous sommes concentrés sur lemodèles de la MHD complète. Ensuite, l'approche numérique est décrite dans lecadre de la stabilisation Variationelle Multi-Échelles (VMS). Cette stabilisationvient ajouter un terme à la formulation faible pour mimer les effets des échellesnon-résolues sur celles résolues. Si les effets de ces sous-échelles ne sont paspris en compte lorsque l'on traite des écoulements dominés par convection,comme dans le cadre des plasmas de fusion, le schéma numérique conduit àdes résultats non-physiques. Une étude détaillée de l'instabilité de « Kinkinterne » a été faite ainsi qu'une étude préliminaire des plasmas avec point-Xayant pour but la validation du schéma numérique développé ici / The main objective of this thesis concerns the implementation of a robuststabilized finite element method for simulating fusion plasmas. For that, we firstderive the magnetohydrodynamic models from the kinetic model. MHD modelsare generally used for macroscopic simulations of plasma instabilities. Weconcentrate ou efforts on the full MHD model. Next, the numerical approach isdescribed in the context of the Variational Multi-Scale (VMS) stabilization. Thisstabilization comes to add a term to the weak formulation to mimics the effectsof the unresolved scales over the coarse scales. If the effects of these subscalesare not taken into account when dealing with fluxes dominated byconvection, as it is the cases for fusion plasmas, the numerical scheme canlead to unphysical results. A detailed study of the resistive internal kinkinstability has been done as well as an introductory study of the so called Xpointplasmas in order to validate the numerical scheme developed here

Cette thèse traite des équations de Maxwell en domaine temporel. Le principal objectif est de proposer des méthodes de type éléments finis d'ordre élevé pour les équations de Maxwell et des schémas d'intégration en temps efficaces sur des maillages localement raffinés. Nous considérons des méthodes GDDT (Galerkine Discontinues en Domaine Temporel) s'appuyant sur une interpolation polynomiale d'ordre arbitrairement élevé des composantes du champ électromagnétique. Les méthodes GDDT pour les équations de Maxwell s'appuient le plus souvent sur des schémas d'intégration en temps explicites dont la condition de stabilité peut être très restrictive pour des maillages raffinés. Pour surmonter cette limitation, nous considérons des schémas en temps qui consistent à appliquer un schéma implicite localement, dans les régions raffinées, tout en préservant un schéma explicite sur le reste du maillage. Nous présentons une étude théorique complète et une comparaison de deux méthodes GDDT localement implicites. Des expériences numériques en 2D et 3D illustrent l'utilité des schémas proposés. Le traitement numérique de milieux de propagation complexes est également l'un des objectifs. Nous considérons l'interaction des ondes électromagnétiques avec les tissus biologiques qui est au cœur de nombreuses applications dans le domaine biomédical. La modélisation numérique nécessite alors de résoudre le système de Maxwell avec des modèles appropriés de dispersion. Nous formulons une méthode GDDT localement implicite pour le modèle de Debye et proposons une analyse théorique et numérique complète du schéma. / This work deals with the time-domain formulation of Maxwell's equations. The main objective is to propose high-order finite element type methods for the discretization of Maxwell's equations and efficient time integration methods on locally refined meshes. We consider Discontinuous Galerkin Time-Domain (DGTD) methods relying on an arbitrary high-order polynomial interpolation of the components of the electromagnetic field. Existing DGTD methods for Maxwell's equations often rely on explicit time integration schemes and are constrained by a stability condition that can be very restrictive on highly refined meshes. To overcome this limitation, we consider time integration schemes that consist in applying an implicit scheme locally i.e. in the refined regions of the mesh, while preserving an explicit scheme in the complementary part. We present a full theoretical study and a comparison of two locally implicit DGTD methods. Numerical experiments for 2D and 3D problems illustrate the usefulness of the proposed time integration schemes. The numerical treatment of complex propagation media is also one of the objectives. We consider the interaction of electromagnetic waves with biological tissues that is of interest to applications in biomedical domain. Numerical modeling then requires to solve the system of Maxwell's equations coupled to appropriate models of physical dispersion. We derive a locally implicit DGTD method for the Debye model and we achieve a full theoretical and numerical analysis of the resulting scheme.