Tato práce se zabývá Runge--Kuttovými metodami pro počáteční problém. Práce začíná analýzou Eulerovy metody a odvozením podmínek řádu. Jsou představeny modifikované metody. Pro dvě z nich je určen jejich řád teoreticky a pro všechny je provedeno numerické testování řádu. Jsou představeny a numericky testovány dva typy metod s odhadem chyby, "embedded" metody a metody založené na modifikovaných metodách. V druhé části jsou odvozeny implicitní metody. Jsou představeny dva způsoby konstrukce implicitních "embedded" metod. Jsou zmíněny také diagonální implicitní metody. Na závěr jsou probrány dva druhy stability u metod prezentovaných v práci.

In this diplomatic work, the issue of low-temperature asphalts and binders for asphalt mixtures is solved. Diploma work is divided into two parts, theoretical and practical. In the theoretical part of diploma work, the purpose was to process a overview of technologies that allow the reduction of working temperature in the production and laying of asphalt mixtures. The purpose of the practical part was to comparison the functional parameters of two types of asphalt mixtures made in two variants, with the common asphalt binder and adding licomont into asphalt. Attention was given to stiffness, wheel drive tests and low temperature parameters.

The solving of stiff systems is still a contemporary sophisticated problem. The basic problem is the absence of precise definition of stiff systems. A question is also how to detect the stiffness in a given system of differential equations. Implicit numerical methods are commonly used for solving stiff systems. The stability domains of these methods are relatively large but the order of them is low.   The thesis deals with numerical solution of ordinary differential equations, especially numerical calculations using Taylor series methods. The source of stiffness is analyzed and the possibility how to reduce stiffness             in systems of ordinary differential equations (ODEs) is introduced. The possibility of detection stiff systems using explicit Taylor series terms is analyzed. The stability domains of explicit and implicit Taylor series are presented. The solutions of stiff systems using implicit Taylor series method are presented in many examples. The multiple arithmetic must be used in many cases. The new suitable parallel algorithm based on implicit Taylor series method with recurrent calculation of Taylor series terms and Newton iteration method (ITMRN) is proposed.

This thesis aims to explore if multibody simulations is a suitable method to investigate vibrations in the steering system of trucks. Vibrations in the steering system and subsequently in the steering wheel is a common issue that automotive manufacturers face. The vibration levels in the steering wheel are in some countries regulated and some vibration phenomena can even cause issues with the handling properties of the whole vehicle. Therefore being able to predict and reduce these with the help of multibody simulations would be of great value. The thesis does this by comparing the simulations to measurements. It investigates what parts can be approximated as rigid, what the effects different numerical solvers have and compares different driving scenarios. This can however be quite challenging, one reason being that the differential equations arising when performing multibody simulations of trucks are very stiff. The numerical challenges of this must be overcome while still keeping the resolution of the accelerations in the solution high enough to still be representative of reality. The thesis also explains how to mathematically model a physical system such that the numerical analysis of it can be efficient. The results show that the success of multibody simulations is very dependent on the test case. However, they also show that together with physical measurements multibody simulations can be a powerful complementary tool. The thesis also presents improvements that could be made to the model as well as certain key areas that need to be studied more in order to align the multibody simulations results with measurements. The multibody simulations software used to perform the calculations and the modelling in the report is Adams developed by Hexagon AB. / Den här rapporten syftar till att ge inblick i om flerkroppssimuleringar kan vara ett användbart verktyg för att undersöka styrsystemsvibrationer i lastbilar. Dessa vibrationer är orsaken till en mängd styrningsproblem samt att rattvibrationer har lagkrav att inte vara för stora. Att kunna förutspå och efterlikna dessa vibrationer med flerkroppssimulering skulle därmed vara till stor fördel. Detta undersöks genom att jämföra simuleringarna med mätdata. Det undersöks vilken påverkan stelkroppsapproximationer av vissa komponenter har, påverkan av olika numeriska integrationmetoder samt steglängder och även olika körningslastfall. Att genomföra flerkroppssimuleringar av lastbilar är dock inte alltid helt enkelt, på grund av differentialekvationernas styva karaktär uppstår ofta konvergensproblem. Ska man sedan använda resultaten för att undersöka styrsystemsvibrationer måste man överkomma dessa konvergensproblem men bibehålla en tillräckligt fin upplösning av resultatet för att resultatet fortfarande ska vara representativt av den fysiska lastbilens dynamiska egenskaper. Rapporten beskriver även hur man kan gå tillväga för att matematiskt modellera ett fysisk system så att det effektivt går att utföra dynamisk analys av det. Resultaten visar att flerkroppssimulering kan vara väldigt beroende på vad körfallet är, med vissa körfall där simuleringar och mätningar stämmer väl överens och andra där detta inte är fallet. På grund av detta kan det vara otillräckligt att endast använda flerkroppssimulering för att utvärdera styrsystemsvibrationer, men resultaten visar att tillsammans med mätdata kan flerkroppssimulering vara ett kraftfullt komplement. I rapporten presenteras även exempel av viktiga komponenter att ta hänsyn till för att bättre kunna simulera styrsystemsvibrationer samt områden där mer forskning har potential att förbättra flerkroppssimuleringar i hänsyn till styrsystemsvibrationer. Mjukvaran som används för att utföra flerkroppssimulering är Adams som utvecklas av Hexagon AB.

In Nonlinear Finite Element Analysis (FEA) applied to structures, displacements at which the tangent stiffness matrix KT becomes singular are called critical points, and correspond to instabilities such as buckling or elastoplastic softening (e.g., necking). Prior to the introduction of Arc Length Methods (ALMs), critical points posed severe computational challenges, which was unfortunate since behavior at instabilities is of great interest as a precursor to structural failure. The original ALM was shown to be capable in some circumstances of continued computation at critical points, but limited success and unattractive features of the formulation were noted and addressed in extensive subsequent research. The widely used Crisfield Cylindrical and Spherical ALMs may be viewed as representing the 'state-of-the-art'. The more recent Stiff Arc Length method, which is attractive on fundamental grounds, was introduced in 2004, but without implementation, benchmarking or performance assessment. The present thesis addresses (a) implementation and (b) performance comparisons for the Crisfield and Stiff methods, using simple benchmarks formulated to incorporate elastoplastic softening. It is seen that, in contrast to the Crisfield methods, the Stiff ALM consistently continues accurate computation at, near and beyond critical points.

The mechanisms that control the lateral response of stiff column-supported shallow foundations, resulting from the application of horizontal load on shallow foundations supported by stiff columns, are uncertain. Stiff columns constructed in soft clayey soil have been used to support retaining walls and in such cases, the lateral thrust applied behind these geotechnical structures is a source of horizontal loading. For seismic events, stiff columns constructed in soft clayey soil have been used to support shallow foundations subjected to horizontal load coming from the upper structure of buildings. Due to its practical applications, it has become important to understand the consequences of subjecting a shallow foundation supported by stiff columns to horizontal load by identifying the factors that control the lateral response of such systems. A series of centrifuge tests were carried out to examine the lateral response of stiff column-supported shallow foundations. The experimental trends suggested that the thickness of the coarse-granular mattress placed above the soil-column composite, called the Load Transfer Platform (LTP), controlled the lateral capacity and the overall lateral response of these systems. A numerical study using the finite element method confirmed the experimental trends. A parametric analysis was conducted with the purpose of investigating the influence of different geometry-based and material-based variables in the lateral response of these systems. The results of the parametric analysis further confirmed the importance of the thickness of the LTP in controlling the lateral response. The parametric results also emphasized the contribution of other variables to this lateral response, and these variables included the undrained shear strength of the soft clayey soil around the stiff columns, the stiff column diameter, and the spacing of the stiff columns after they are constructed in the soft clayey soil. / Doctor of Philosophy / Ground improvement is the process of improving the properties of weak soils. In practice, there are several ways to accomplish the ground improvement of weak soils. One way is to use stiff columns. Stiff columns are solid cylinders that are constructed in the weak soil in order to produce a stiff and strong soil-column composite capable of better supporting the square-shaped foundations of structures. Under horizontal load, there is uncertainty on the factors that control the lateral behavior of the stiff columns when used for the support of the square-shaped foundations of structures. An experimental and numerical approach was used to determine these factors and to understand their influence in the lateral behavior of such systems. The findings showed that the controlling factors of this lateral behavior included the thickness of a sand layer placed above the stiff columns, the diameter of the stiff columns, the spacing of the stiff columns after they are constructed in the weak soil, and the strength of the soil around the stiff columns. These findings will aid in improving the design of stiff columns used for the support of square-shaped foundations of structures subjected to horizontal load.

This thesis develops the methodology for solving initial boundary value problems with the use of summation-by-parts discretizations. The combination of high orders of accuracy and a systematic approach to construct provably stable boundary and interface procedures makes this methodology especially suitable for scientific computations with high demands on efficiency and robustness. Most classes of high order methods can be applied in a way that satisfies a summation-by-parts rule. These include, but are not limited to, finite difference, spectral and nodal discontinuous Galerkin methods. In the first part of this thesis, the summation-by-parts methodology is extended to the time domain, enabling fully discrete formulations with superior stability properties. The resulting time discretization technique is closely related to fully implicit Runge-Kutta methods, and may alternatively be formulated as either a global method or as a family of multi-stage methods. Both first and second order derivatives in time are considered. In the latter case also including mixed initial and boundary conditions (i.e. conditions involving derivatives in both space and time). The second part of the thesis deals with summation-by-parts discretizations on multi-block and hybrid meshes. A new formulation of general multi-block couplings in several dimensions is presented and analyzed. It collects all multi-block, multi-element and  hybrid summation-by-parts schemes into a single compact framework. The new framework includes a generalized description of non-conforming interfaces based on so called summation-by-parts preserving interpolation operators, for which a new theoretical accuracy result is presented.

Drilled shaft retaining walls are common earth retaining structures, well suited to urban environments where noise, space, and damage to adjacent structures are major considerations. The design of drilled shaft retaining walls in non-expansive soils is well established. In expansive soils, however, there is no consensus on the correct way to account for the influence of soil expansion on wall behavior. Based on the range of design assumptions currently in practice, existing walls could be substantially over- or under-designed. The goal of this research is to advance the understanding of the effects of expansive clay on drilled shaft retaining walls. The main objectives of this study are to identify the processes responsible for wall loading and deformation in expansive clay, to evaluate how these processes change with time, and to provide guidance for design practice to account for these processes and ensure adequate wall performance. The primary source of information for this research is performance data from a four-year monitoring program at the Lymon C. Reese research wall, a full-scale instrumented drilled shaft retaining wall constructed through expansive clay in Manor, Texas. The test wall was instrumented with inclinometers and fiber optic strain gauges, and performance data was recorded during construction, excavation, during natural moisture fluctuations, and during controlled inundation tests that provided the retained soil with unlimited access to water. In addition to the test wall study, a field assessment of existing TxDOT drilled shaft retaining walls was conducted. The main process influencing short-term wall deformation was found to be global response to stress relief during excavation, which causes the wall and soil to move together without the development of large earth pressures or bending stresses. Long-term wall deformations were governed by the development of drained conditions in both the retained soil and the foundation soil after approximately eight months of controlled inundation testing. To ensure adequate wall performance, the deformations and structural loads associated with short- and long-term conditions should be combined and checked against allowable values. / text

Les simulations numériques réalistes en électrophysiologie cardiaque ont un coût de calcul extrêmement élevé. Ce coût s’explique en grande partie par la raideur, à la fois en temps et en espace, d’une onde de « potentiel d’action » (PA). Par ailleurs, les phénomènes observés sont très instationnaires et s’étudient en temps long. Une description précise de la dynamique des PA est cruciale pour construire des modèles numériques pertinents d’un point de vue médical ou clinique. Cet aspect fondamental ne peut être contourné dans les études numériques réalistes.La raideur de l’onde de PA ne peut être captée numériquement qu’en ayant recours à des maillages très fins. Ces maillages très fins induisent un coût de calcul très important, et introduisent aussi des erreurs supplémentaires : les systèmes linéaires à résoudre deviennent très mal conditionnés. Au final, les erreurs numériques peuvent être particulièrement grandes dans les simulations alors que leur contrôle est évidemment essentiel pour assurer la fiabilité des résultats. Jusqu’à présent, très peu de résultats sont disponibles pour assurer cette fiabilité. Dans les faits, les erreurs sont la plupart du temps contrôlées par des procédés empiriques. Il existe quelques résultats théoriques étudiant la convergence et la stabilité des schémas numériques associés. En pratique, en plus d'avoir un contrôle de l'erreur sur le potentiel, il est aussi nécessaire d'avoir un contrôle de l’erreur sur des quantités macroscopiques décrivant la dynamique de l’onde de PA : temps d’activation, durée du PA, propriétés de restitution... Ces quantités ont en effet une interprétation physiologique qui permet de caractériser le caractère arythmogène des tissus.Les modèles sont des systèmes d’EDP de réaction-diffusion couplés avec des systèmes d’équations différentielles pouvant être très raides, les modèles ioniques. Ils sont actuellement discrétisés par éléments finis conforme (Lagrange) et par des schémas en temps d’ordre un ou deux. Dans ce travail, nous concevons et évaluons l’intérêt d'utiliser des méthodes d’ordre supérieure pour ces systèmes. Parallèlement nous introduisons d'une part une nouvelle classe de schémas appelé schémas exponentiel Adams Bashforth intégral (IEAB), et d'autre part des schémas Rush Larsen (RL) d'ordre élevé. Ces nouveaux schémas sont des schémas multipas de type exponentiels. Nous montrons qu'ils possèdent des bonnes propriétés de stabilité et permettent de faire face efficacement à la raideur des modèles ioniques. Les schémas que nous proposons sont comparés numériquement (en terme de précision, coût en temps de calcul et stabilité) à plusieurs schémas classiques, ainsi qu'aux schémas exponentiels (RL1, RL2) communément utilisés pour des simulations en électrophysiologie cardiaque. Nous proposons des techniques permettant de calculer avec précision les quantités d’intérêts cliniques (temps d’activation, de récupération, durée du potentiel d’action). Des résultats théoriques de convergence en temps et de convergence globale (espace et temps) sont énoncés et prouvés. Ces résultats sont ensuite illustrés numériquement à travers le modèle monodomaine et les modèles ioniques de Beeler Reuter, de Ten Tusscher et al. L’intérêt d'utiliser des schémas d'ordre élevés est aussi évalué sur des ondes spirales en 2D et 3D. / Realistic numerical simulations in cardiac electrophysiology have a computational cost of extremely high. This cost is largely explained by the stiffness both in time and space, of the action potential (AP) wave. Moreover, the observed phenomena are very unsteady and are studied in long time. A precise description of the dynamic of AP is crucial for constructing relevant numerical models, from a medical or clinical perspective. This fundamental aspect can not be circumvented in realistic numerical studies.The stiffness of AP wave can only be captured numerically, by using very fine meshes. In addition to the high computational cost, these very fine meshes also introduce additional errors : the linear systems to solve become very badly conditioned. In the end, the numerical errors can be particularly large whereas their control is obviously essential to ensure the reliability of the results. So far very few results are available to ensure this reliability. In practice, the errors are mostly controlled by empirical processes. In practice, in addition of having a control of the error on the potential, it is also necessary to have an error control on macroscopic quantities describing the dynamics of the AP wave : activation time, AP duration, properties of restitution ... These quantities have indeed a physiological interpretation which allows to characterize the arrhythmogenic character of the tissues.The models are systems of reaction diffusion PDE coupled with systems of differential equations that can be very stiffs (ionic models). They are currently discretized by conforming finite elements (Lagrange finite elements methods) and by schemes in time of order one or two. In this work, we design and evaluate the interest of using higher order methods for these systems. At the same time, we introduce on the one hand, a new class of schemes called Integral Exponential Adams Bashforth (IEAB) schemes and, on the other hand, high order Rush Larsen (RL) schemes. These new schemes are exponential time-stepping schemes. We show that they have good stability properties and can efficiently cope with the stiffness of ionic models. The schemes we propose are numerically compared (in terms of accuracy, CPU time and stability) with several classical schemes, as well as with the exponential schemes (RL1, RL2), commonly used for cardiac electrophysiology simulations. We propose good techniques for accurately calculating quantities of clinical interest (activation time, recovery time, duration of action potential). Theoretical results of convergence in time and global convergence (in space and time) are stated and proved. These results are then illustrated numerically through the monodomain model and the ionic models of Beeler Reuter, Ten Tusscher et al. The advantage of using high order schemes is also evaluated on spiral waves in 2D and 3D.

No description available.