Context. In computer graphics, realistic looking fluid is often desired. Simulating realistic fluids is a time consuming and computationally expensive task, therefore, much research has been devoted to reducing the simulation time while maintaining the realism. Two of the more recent optimization algorithms within particle based simulations are two-scale simulation and regional time stepping (RTS). Both of them are based on the predictive-corrective incompressible smoothed particle hydrodynamics (PCISPH) algorithm. Objectives. These algorithms improve on two separate aspects of PCISPH, two-scale simulation reduces the number of particles and RTS focuses computational power on regions of the fluid where it is most needed. In this paper we have developed and investigated the performance of an algorithm combining them, utilizing both optimizations. Methods. We implemented both of the base algorithms, as well as PCISPH, before combining them. Therefore we had equal conditions for all algorithms when we performed our experiments, which consisted of measuring the time it took to run each algorithm in three different scene configurations. Results. Results showed that our combined algorithm on average was faster than the other three algorithms. However, our implementation of two-scale simulation gave results inconsistent with the original paper, showing a slower time than even PCISPH. This invalidates the results for our combined algorithm since it utilizes the same implementation. Conclusions. We see that our combined algorithm has potential to speed up fluid simulations, but since the two-scale implementation was incorrect, our results are inconclusive.

Elliptic obstacle problems are formulated to find either superharmonic solutions or minimal surfaces that lie on or over the obstacles, by incorporating inequality constraints. This dissertation investigates simple iterative algorithms based on the successive over-relaxation (SOR) method. It introduces subgrid methods to reduce accuracy deterioration occurring near the free boundary when the mesh grid does not match with the free boundary. For nonlinear obstacle problems, a method of gradient-weighting is introduced to solve the problem more conveniently and efficiently. The iterative algorithm is analyzed for convergence for both linear and nonlinear obstacle problems. Parabolic initial-boundary value problems with nonsmooth data show either rapid transitions or reduced smoothness in its solution. For those problems, specific numerical methods are required to avoid spurious oscillations as well as unrealistic smoothing of steep changes in the numerical solution. This dissertation investigates characteristics of the θ-method and introduces a variable-θ method as a synergistic combination of the Crank-Nicolson (CN) method and the implicit method. It suppresses spurious oscillations, by evolving the solution implicitly at points where the solution shows a certain portent of oscillations or reduced smoothness, and maintains as a similar accuracy as the CN method with smooth data. An effective strategy is suggested for the detection of points where the solution may introduce spurious oscillations (the wobble set); the resulting variable-θ method is analyzed for its accuracy and stability. After a theory of morphogenesis in chemical cells was introduced in 1950s, much attention had been devoted to the numerical solution of reaction-diffusion (RD) equations. This dissertation studies a nonoscillatory second-order time-stepping procedure for RD equations incorporating with variable-θ method, as a perturbation of the CN method. We also perform a sensitivity analysis for the numerical solution of RD systems to conclude that it is much more sensitive to the spatial mesh resolution than the temporal one. Moreover, to enhance the spatial approximation of RD equations, this dissertation investigates the averaging scheme, that is, an interpolation of the standard and skewed discrete Laplacian operator and introduce the simple optimizing strategy to minimize the leading truncation error of the scheme.

The electrical activity of the heart is a well studied process. Mathematical modeling and computer simulations are used to study the cardiac electrical activity: several mathematical models exist, among them the microscopic model, which is based on the explicit representation of individual cells. The cardiac tissue is viewed as two separate domains: the intra-cellular and extra-cellular domains, Ωᵢ and Ωₑ, respectively, separated by cellular membranes Γ. The microscopic model consists of a set of Poisson equations, one for each sub-domain, Ωᵢ and Ωₑ, coupled on interfaces Γ with nonlinear transmission conditions involving a system of ODEs. The unusual transmission conditions on Γ make the model challenging to solve numerically. In this thesis, we first focus on the dimensional analysis of the microscopic model. We then reformulate the problem on the interface Γ using a Steklov-Poincaré operator. We discretize the model in space using finite element methods. We prove the existence of a semi-discrete solution using a reformulation of the model as an ODE system on the interface Γ. We derive stability and error estimates for the finite element method. Afterwards, we consider five numerical schemes including the Godunov splitting method, two implicit methods, (Backward Euler (BE) and second order Backward Differentiation Formula (BDF2)), and two semi-implicit methods (Forward Backward Euler (FBE), and second order Semi-implicit Backward Differentiation Formula (SBDF2)). A convergence analysis of the implicit and semi-implicit methods is performed and the results are compared with manufactured solutions that we have proposed. Numerical results are presented to compare the stability, accuracy and efficiency of the methods. CPU times needed to solve the problem over a single cell using FBE, SBDF2 and Godunov splitting methods are reported. The results show that FBE and Godunov splitting methods achieve better numerical accuracy and efficiency than implicit and SBDF2 schemes, for a given computational time. Finally, we solve the model using FBE and Domain Decomposition Method (DDM) for two cells connected to each other by a gap junction. We investigate the influence of the space discretization and we explore the differences between a conforming and nonconforming mesh on Γ. We compare the solutions obtained with both FBE and DDM methods. The results show that both methods give the same solution. Therefore, the DDM is capable of providing an accurate solution with a minimal number of sub-domain iterations.

Air Quality Models are defined for prediction and simulation of air pollutant concentrations over a certain period of time. The predictions can be used in setting limits for the emission levels of industrial facilities. The input data for the air quality models are very large and encompass various environmental conditions like wind speed, turbulence, temperature and cloud density. Most air quality models are based on advection-diffusion equations. These differential equations are moderately stiff and require appropriate techniques for fast integration over large intervals of time. Implicit time stepping techniques for solving differential equations being unconditionally stable are considered suitable for the solution. However, implicit time stepping techniques impose certain data dependencies that can cause the parallelization of air quality models to be inefficient. The current approach uses Runge Kutta Chebyshev explicit method for solution of advection diffusion equations. It is found that even if the explicit method used is computationally more expensive in the serial execution, it takes lesser execution time when parallelized because of less complicated data dependencies presented by the explicit time-stepping. The implicit time-stepping on the other hand cannot be parallelized efficiently because of the inherent complicated data dependencies. / Master of Science

This thesis investigates pseudo time stepping as an implicit solver for non-uniform meshes. Specifically, on meshes with small cells. Regions with small cells can be treated with a time stepping scheme known as mixed explicit implicit time stepping, which treats small cells implicitly to ensure stability. The goal of this thesis is to examine whether pseudo time stepping is a good solver for the implicit part. To investigate this, results from pseudo time stepping on a uniform mesh are first compared with a direct implicit solver. This is done to ensure that both methods produce the same results. Then, tests are conducted on a non-uniform mesh, featuring one small cell in the center of the domain. The aim of these tests is to evaluate how the parameters in pseudo time stepping contribute to the accuracy and number of iterations in a small mesh. With respect to this, the thesis aims to answer whether and when pseudo time stepping is a feasible solver. The results show promising potential in pseudo time stepping, as similar results to the direct solver on the uniform mesh are achieved with few iterations. It also shows feasible iteration numbers for the non-equidistant case, nearly regardless of number of grid points or the size of the small cell.

Time integration methods are necessary for the solution of transient flow problems. In recent years, interest in transient flow problems has increased, leading to a need for better understanding of the costs and benefits of various time integration schemes. The present work investigates two common time integration schemes, namely the Semi-Implicit Method for Pressure-Linked Equations (SIMPLE) and the Fractional Step (FS) method. Three two-dimensional, transient, incompressible flow problems are solved using a cell centered, finite volume code. The three test cases are laminar flow in a lid-driven skewed cavity, laminar flow over a square cylinder, and turbulent flow over a square cylinder. Turbulence is modeled using wall functions and the k - ε turbulence model with the modifications suggested by Kato and Launder. Solution efficiency as measured by the effort carried out by the flow equation solver and CPU time is examined. Accuracy of the results, generated using the SIMPLE and FS time integration schemes, is analyzed through a comparison of the results with existing experimental and/or numerical solutions. Both the SIMPLE and FS algorithms are shown to be capable of solving benchmark flow problems with reasonable accuracy. The two schemes differ slightly in their prediction of flow evolution over time, especially when simulating very slowly changing flows. As the time step size decreases, the SIMPLE algorithm computational cost (CPU time) per time step remains approximately constant, while the FS method experiences a reduction in cost per time step. Also, the SIMPLE algorithm is numerically stable for time steps approaching infinity, while the FS scheme suffers from numerical instability if the time step size is too large. As a result, the SIMPLE algorithm is recommended to be used for transient simulations with large time steps or steady state problems while the FS scheme is better suited for small time step solutions, although both time-stepping schemes are found to be most efficient when their time steps are at their maximum stable value.

Time integration methods are necessary for the solution of transient flow problems. In recent years, interest in transient flow problems has increased, leading to a need for better understanding of the costs and benefits of various time integration schemes. The present work investigates two common time integration schemes, namely the Semi-Implicit Method for Pressure-Linked Equations (SIMPLE) and the Fractional Step (FS) method. Three two-dimensional, transient, incompressible flow problems are solved using a cell centered, finite volume code. The three test cases are laminar flow in a lid-driven skewed cavity, laminar flow over a square cylinder, and turbulent flow over a square cylinder. Turbulence is modeled using wall functions and the k - ε turbulence model with the modifications suggested by Kato and Launder. Solution efficiency as measured by the effort carried out by the flow equation solver and CPU time is examined. Accuracy of the results, generated using the SIMPLE and FS time integration schemes, is analyzed through a comparison of the results with existing experimental and/or numerical solutions. Both the SIMPLE and FS algorithms are shown to be capable of solving benchmark flow problems with reasonable accuracy. The two schemes differ slightly in their prediction of flow evolution over time, especially when simulating very slowly changing flows. As the time step size decreases, the SIMPLE algorithm computational cost (CPU time) per time step remains approximately constant, while the FS method experiences a reduction in cost per time step. Also, the SIMPLE algorithm is numerically stable for time steps approaching infinity, while the FS scheme suffers from numerical instability if the time step size is too large. As a result, the SIMPLE algorithm is recommended to be used for transient simulations with large time steps or steady state problems while the FS scheme is better suited for small time step solutions, although both time-stepping schemes are found to be most efficient when their time steps are at their maximum stable value.

The purpose of this thesis is to develop corotational beam elements for the nonlinear dynamic analyse of flexible beam structures. Whereas corotational beam elements in statics are well documented, the derivation of a corotational dynamic formulation is still an issue. In the first journal paper, an efficient dynamic corotational beam formulation is proposed for 2D analysis. The idea is to adopt the same corotational kinematic description in static and dynamic parts. The main novelty is to use cubic interpolations to derive both inertia terms and internal terms in order to capture correctly all inertia effects. This new formulation is compared with two classic formulations using constant Timoshenko and constant lumped mass matrices. In the second journal paper, several choices of parametrization and several time stepping methods are compared. To do so, four dynamic formulations are investigated. The corotational method is used to develop expressions of the internal terms, while the dynamic terms are formulated into a total Lagrangian context. Theoretical derivations as well as practical implementations are given in detail. Their numerical accuracy and computational efficiency are then compared. Moreover, four predictors and various possibilities to simplify the tangent inertia matrix are tested. In the third journal paper, a new consistent beam formulation is developed for 3D analysis. The novelty of the formulation lies in the use of the corotational framework to derive not only the internal force vector and the tangent stiffness matrix but also the inertia force vector and the tangent dynamic matrix. Cubic interpolations are adopted to formulate both inertia and internal local terms. In the derivation of the dynamic terms, an approximation for the local rotations is introduced and a concise expression for the global inertia force vector is obtained. Four numerical examples are considered to assess the performance of the new formulation against two other ones based on linear interpolations. Finally, in the fourth journal paper, the previous 3D corotational beam element is extended for the nonlinear dynamics of structures with thin-walled cross-section by introducing the warping deformations and the eccentricity of the shear center. This leads to additional terms in the expressions of the inertia force vector and the tangent dynamic matrix. The element has seven degrees of freedom at each node and cubic shape functions are used to interpolate local transversal displacements and axial rotations. The performance of the formulation is assessed through five examples and comparisons with Abaqus 3D-solid analyses. / <p>QC 20131017</p>

Computational fluid dynamics (CFD) is a significant tool routinely used indesign and optimization in aerospace industry. Often cases with unsteadyflows must be computed, and the long compute times of standard methods hasmotivated the present work on new implicit methods to replace the standardexplicit schemes. The implementation and numerical experiments were donewith the Swedish national flow solver Edge, developed by FOI,universities, and collaboration partners.The work is concentrated on a Lower-Upper Symmetric Gauss-Seidel (LU-SGS)type of time stepping. For the very anisotropic grids needed forReynolds-Averaged Navier-Stokes (RANS) computations of turbulent boundary layers,LU-SGS is combined with a line-implicit technique.  The inviscid flux Jacobians which contribute to the diagonalblocks of the system matrix are based on a flux splitting method with upwind type dissipation giving  control over diagonal dominance and artificial dissipation.The method is  controlled by several parameters, and comprehensivenumerical experiments were carried out to identify their influence andinteraction so that close to optimal values can be suggested. As an example,the optimal number of iterations carried out in a time-step increases with increased resolution of the computational grid.The numbering of the unknowns is important, and the numberings produced by mesh generators of Delaunay- and advancing front-type wereamong the best.The solver has been parallelized with the Message Passing Interface (MPI) for runs on multi-processor hardware,and its performance scales with the number of processors at least asefficiently as the explicit methods. The new method saves typicallybetween 50 and 80 percent of the runtime, depending on the case, andthe largest computations have reached 110M grid nodes. Theclassical multigrid acceleration for 3D RANS simulations was foundineffective in the cases tested in combination with the LU-SGS solverusing optimal parameters. Finally, preliminary time-accurate simulations for unsteady flows have shown promising results. / <p>QC 20141201</p>

Du fait de sa robustesse et de son faible coût, la machine synchro-réluctante (MSR) constitue une alternative intéressante à la machine asynchrone. A pertes égales, une MSR bien optimisée offre un couple et par suite un rendement plus élevés. Ainsi, la MSR est très compétitive pour les applications à haute vitesse, à forte puissance ou à haute température. Cette thèse se propose d’optimiser les machines synchro-réluctantes à rotor massif et avec barrières de flux pour produire le maximum de couple avec un facteur de puissance le plus élevé possible. Pour cela, une modélisation originale utilisant des réseaux de perméances non linéaires a été mise au point pour les deux types de MSR. Les modèles proposés sont significativement plus rapides et aussi précis que les modèles par éléments finis. De plus, la réalisation d’un prototype à barrières de flux a permis de les valider expérimentalement. / Because of its robustness and its low cost, the synchronous reluctance motor (SynRM) is an interesting alternative to the induction motor. At equal losses, a correctly optimized SynRM offers a higher torque and then a higher efficiency. Thus, the SynRM is very comptetitive for high speed, high power or high temperature applications. This thesis intends to optimize massive rotor and flux barrier rotor SynRM to produce the maximum torque with the highest possible power factor. For this purpose, an original non linear reluctance network modeling of synchronous reluctance motors with a massive or a flux barrier rotor was developed. The proposed models are significantly faster than the finite element ones and take accurately into account the saturation of all ferromagnetic parts of the motor. The construction of a flux barrier rotor prototype allowed an experimental validation of the modeling approach.