Infimalfolgen nichtkonvexer Variationsprobleme haben aufgrund feiner Oszillationen häufig keinen starken Grenzwert in Sobolevräumen. Diese Oszillationen haben eine physikalische Bedeutung; Finite-Element-Approximationen können sie jedoch im Allgemeinen nicht auflösen. Relaxationsmethoden ersetzen die nichtkonvexe Energie durch ihre (semi)konvexe Hülle. Das entstehende makroskopische Modell ist degeneriert: es ist nicht strikt konvex und hat eventuell mehrere Minimalstellen. Die fehlende Kontrolle der primalen Variablen führt zu Schwierigkeiten bei der a priori und a posteriori Fehlerschätzung, wie der Zuverlässigkeits- Effizienz-Lücke und fehlender starker Konvergenz. Zur Überwindung dieser Schwierigkeiten erweitern Stabilisierungstechniken die relaxierte Energie um einen diskreten, positiv definiten Term. Bartels et al. (IFB, 2004) wenden Stabilisierung auf zweidimensionale Probleme an und beweisen dabei starke Konvergenz der Gradienten. Dieses Ergebnis ist auf glatte Lösungen und quasi-uniforme Netze beschränkt, was adaptive Netzverfeinerungen ausschließt. Die vorliegende Arbeit behandelt einen modifizierten Stabilisierungsterm und beweist auf unstrukturierten Netzen sowohl Konvergenz der Spannungstensoren, als auch starke Konvergenz der Gradienten für glatte Lösungen. Ferner wird der sogenannte Fluss-Fehlerschätzer hergeleitet und dessen Zuverlässigkeit und Effizienz gezeigt. Für Interface-Probleme mit stückweise glatter Lösung wird eine Verfeinerung des Fehlerschätzers entwickelt, die den Fehler der primalen Variablen und ihres Gradienten beschränkt und so starke Konvergenz der Gradienten sichert. Der verfeinerte Fehlerschätzer konvergiert schneller als der Fluss- Fehlerschätzer, und verringert so die Zuverlässigkeits-Effizienz-Lücke. Numerische Experimente mit fünf Benchmark-Tests der Mikrostruktursimulation und Topologieoptimierung ergänzen und bestätigen die theoretischen Ergebnisse. / Infimising sequences of nonconvex variational problems often do not converge strongly in Sobolev spaces due to fine oscillations. These oscillations are physically meaningful; finite element approximations, however, fail to resolve them in general. Relaxation methods replace the nonconvex energy with its (semi)convex hull. This leads to a macroscopic model which is degenerate in the sense that it is not strictly convex and possibly admits multiple minimisers. The lack of control on the primal variable leads to difficulties in the a priori and a posteriori finite element error analysis, such as the reliability-efficiency gap and no strong convergence. To overcome these difficulties, stabilisation techniques add a discrete positive definite term to the relaxed energy. Bartels et al. (IFB, 2004) apply stabilisation to two-dimensional problems and thereby prove strong convergence of gradients. This result is restricted to smooth solutions and quasi-uniform meshes, which prohibit adaptive mesh refinements. This thesis concerns a modified stabilisation term and proves convergence of the stress and, for smooth solutions, strong convergence of gradients, even on unstructured meshes. Furthermore, the thesis derives the so-called flux error estimator and proves its reliability and efficiency. For interface problems with piecewise smooth solutions, a refined version of this error estimator is developed, which provides control of the error of the primal variable and its gradient and thus yields strong convergence of gradients. The refined error estimator converges faster than the flux error estimator and therefore narrows the reliability-efficiency gap. Numerical experiments with five benchmark examples from computational microstructure and topology optimisation complement and confirm the theoretical results.

The main body of this thesis consists of six appended papers. In the  first two, different  cooperative surveillance problems are considered. The second two consider different aspects of the trajectory planning problem, while the last two deal with observer design for mobile robotic and Euler-Lagrange systems respectively.In Papers A and B,  a combinatorial optimization based framework to cooperative surveillance missions using multiple Unmanned Ground Vehicles (UGVs) is proposed. In particular, Paper A  considers the the Minimum Time UGV Surveillance Problem (MTUSP) while Paper B treats the Connectivity Constrained UGV Surveillance Problem (CUSP). The minimum time formulation is the following. Given a set of surveillance UGVs and a polyhedral area, find waypoint-paths for all UGVs such that every point of the area is visible from  a point on a waypoint-path and such that the time for executing the search in parallel is minimized.  The connectivity constrained formulation  extends the MTUSP by additionally requiring the induced information graph to be  kept recurrently connected  at the time instants when the UGVs  perform the surveillance mission.  In these two papers, the NP-hardness of  both these problems are shown and decomposition techniques are proposed that allow us to find an approximative solution efficiently in an algorithmic manner.Paper C addresses the problem of designing a real time, high performance trajectory planner for an aerial vehicle that uses information about terrain and enemy threats, to fly low and avoid radar exposure on the way to a given target. The high-level framework augments Receding Horizon Control (RHC) with a graph based terminal cost that captures the global characteristics of the environment.  An important issue with RHC is to make sure that the greedy, short term optimization does not lead to long term problems, which in our case boils down to two things: not getting into situations where a collision is unavoidable, and making sure that the destination is actually reached. Hence, the main contribution of this paper is to present a trajectory planner with provable safety and task completion properties. Direct methods for trajectory optimization are traditionally based on a priori temporal discretization and collocation methods. In Paper D, the problem of adaptive node distribution is formulated as a constrained optimization problem, which is to be included in the underlying nonlinear mathematical programming problem. The benefits of utilizing the suggested method for  online  trajectory optimization are illustrated by a missile guidance example.In Paper E, the problem of active observer design for an important class of non-uniformly observable systems, namely mobile robotic systems, is considered. The set of feasible configurations and the set of output flow equivalent states are defined. It is shown that the inter-relation between these two sets may serve as the basis for design of active observers. The proposed observer design methodology is illustrated by considering a  unicycle robot model, equipped with a set of range-measuring sensors. Finally, in Paper F, a geometrically intrinsic observer for Euler-Lagrange systems is defined and analyzed. This observer is a generalization of the observer proposed by Aghannan and Rouchon. Their contractivity result is reproduced and complemented  by  a proof  that the region of contraction is infinitely thin. Moreover, assuming a priori bounds on the velocities, convergence of the observer is shown by means of Lyapunov's direct method in the case of configuration manifolds with constant curvature. / QC 20100622 / TAIS, AURES

<p>The main body of this thesis consists of four appended papers. The first two consider different aspects of the trajectory planning problem, while the last two deal with observer design for mobile robotic and Euler-Lagrange systems respectively.</p><p>The first paper addresses the problem of designing a real time, high performance trajectory planner for aerial vehicles. The main contribution is two-fold. Firstly, by augmenting a novel safety maneuver at the end of the planned trajectory, this paper extends previous results by having provable safety properties in a 3D setting. Secondly, assuming initial feasibility, the planning method is shown to have finite time task completion. Moreover, in the second part of the paper, the problem of simultaneous arrival of multiple aerial vehicles is considered. By using a time-scale separation principle, one is able to adopt standard Laplacian control to this consensus problem, which is neither unconstrained, nor first order.</p><p>Direct methods for trajectory optimization are traditionally based on<i> a</i> <i>priori </i>temporal discretization and collocation methods. In the second paper, the problem of adaptive node distribution is formulated as a constrained optimization problem, which is to be included in the underlying nonlinear mathematical programming problem. The benefits of utilizing the suggested method for online trajectory optimization are illustrated by a missile guidance example.</p><p>In the third paper, the problem of active observer design for an important class of non-uniformly observable systems, namely mobile robotics systems, is considered. The set of feasible configurations and the set of output flow equivalent states are defined. It is shown that the inter-relation between these two sets may serve as the basis for design of active observers. The proposed observer design methodology is illustrated by considering a unicycle robot model, equipped with a set of range-measuring sensors.</p><p>Finally, in the fourth paper, a geometrically intrinsic observer for Euler-Lagrange systems is defined and analyzed. This observer is a generalization of the observer recently proposed by Aghannan and Rouchon. Their contractivity result is reproduced and complemented by a proof that the region of contraction is infinitely thin. However, assuming <i>a</i> <i>priori </i>bounds on the velocities, convergence of the observer is shown by means of Lyapunov's direct method in the case of configuration manifolds with constant curvature.</p>

In power plants using large utility coal-fired boilers for generation of electricity the coal is pulverised in coal mills and then it has to be pneumatically transported and distributed to a larger number of burners (e.g. 30-40) circumferentially arranged in several rows around the burning chamber of the boiler. Besides the large pipework flow splitting devices are necessary for distribution of an equal amount of pulverised fuel (PF) to each of the burners. So called trifurcators (without inner fittings or guiding vanes) and ''riffle'' type bifurcators are commonly used to split the gas-coal particle flow into two or three pipes/channels with an equal amount of PF mass flow rate in each outflow cross section of the flow splitting device. These PF flow splitting devices are subject of a number of problems. First of all an uneven distribution of PF over the burners of a large utility boiler leads to operational and maintenance problems, increased level of unburned carbon and higher rates of NOX emissions. Maldistribution of fuel between burners caused by non uniform concentration of the PF (particle roping) in pipe and channel bends prior to flow splitting devices leads to uncontrolled differences in the fuel to air ratio between burners. This results in localised regions in the furnace which are fuel rich, where insufficient air causes incomplete combustion of the fuel. Other regions in the furnace become fuel lean, forming high local concentrations of NOX due to the high local concentrations of O2. Otherwise PF maldistribution can impact on power plant maintenance in terms of uneven wear on PF pipework, flow splitters as well as the effects on boiler panels (PF deposition, corrosion, slagging). In order to address these problems in establishing uniform PF distribution over the outlet cross sections of flow splitting devices in the pipework of coal-fired power plants the present paper deals with numerical prediction and analysis of the complex gas and coal particle (PF) flow through trifurcators and ''riffle'' type bifurcators. The numerical investigation is based on a 3-dimensional Eulerian- Lagrangian approach (MISTRAL/PartFlow-3D) developed by Frank et al. The numerical method is capable to predict isothermal, incompressible, steady gas- particle flows in 3-dimensional, geometrically complex flow geometries using boundary fitted, block-structured, numerical grids. Due to the very high numerical effort of the investigated gas-particle flows the numerical approach has been developed with special emphasis on efficient parallel computing on clusters of workstations or other high performance computing architectures. Besides the aerodynamically interaction between the carrier fluid phase and the PF particles the gas-particle flow is mainly influenced by particle-wall interactions with the outer wall boundaries and the inner fittings and guiding vanes of the investigated flow splitting devices. In order to allow accurate quantitative prediction of the motion of the disperse phase the numerical model requires detailed information about the particle-wall collision process. In commonly used physical models of the particle-wall interaction this is the knowledge or experimental prediction of the restitution coefficients (dynamic friction coefficient, coefficient of restitution) for the used combination of particle and wall material, e.g. PF particles on steel. In the present investigation these parameters of the particle-wall interaction model have been obtained from special experiments in two test facilities. Basic experiments to clarify the details of the particle-wall interaction process were made in a test facility with a spherical disk accelerator. This test facility furthermore provides the opportunity to investigate the bouncing process under normal pressure as well as under vacuum conditions, thus excluding aerodynamically influences on the motion of small particles in the near vicinity of solid wall surfaces (especially under small angles of attack). In this experiments spherical glass beads were used as particle material. In a second test facility we have investigated the real impact of non-spherical pulverised fuel particles on a steel/ceramic target. In this experiments PF particles were accelerated by an injector using inert gas like e.g. CO2 or N2 as the carrier phase in order to avoid dust explosion hazards. The obtained data for the particle-wall collision models were compared to those obtained for glass spheres, where bouncing models are proofed to be valid. Furthermore the second test facility was used to obtain particle erosion rates for PF particles on steel targets as a function of impact angles and velocities. The results of experimental investigations has been incorporated into the numerical model. Hereafter the numerical approach MISTRAL/PartFlow-3D has been applied to the PF flow through a ''riffle'' type bifurcator. Using ICEM/CFD-Hexa as grid generator a numerical mesh with approximately 4 million grid cells has been designed for approximation of the complex geometry of the flow splitting device with all its interior fittings and guiding vanes. Based on a predicted gas flow field a large number of PF particles are tracked throughout the flow geometry of the flow-splitter. Besides mean quantities of the particle flow field like e.g. local particle concentrations, mean particle velocities, distribution of mean particle diameter, etc. it is now possible to obtain information about particle erosion on riffle plates and guiding vanes of the flow splitting device. Furthermore the influence of different roping patterns in front of the flow splitter on the uniformness of PF mass flow rate splitting after the bifurcator has been investigated numerically. Results show the efficient operation of the investigated bifurcator in absence of particle roping, this means under conditions of an uniform PF particle concentration distribution in the inflow cross section of the bifurcator. If particle roping occurs and particle concentration differs over the pipe cross section in front of the bifurcator the equal PF particle mass flow rate splitting can be strongly deteriorated in dependence on the location and intensity of the particle rope or particle concentration irregularities. The presented results show the importance of further development of efficient rope splitting devices for applications in coal-fired power plants. Numerical analysis can be used as an efficient tool for their investigation and further optimisation under various operating and flow conditions.

The transport or the separation of solid particles or droplets suspended in a fluid flow is a common task in mechanical and process engineering. To improve machinery and physical processes (e.g. for coal combustion, reduction of NO_x and soot) an optimization of complex phenomena by simulation applying the fundamental conservation equations is required. Fluid-particle flows are characterized by the ratio of density of the two phases gamma=rho_P/rho_F, by the Stokes number St=tau_P/tau_F and by the loading in terms of void and mass fraction. Those numbers (Stokes number, gamma) define the flow regime and which relevant forces are acting on the particle. Dependent on the geometrical configuration the particle-wall interaction might have a heavy impact on the mean flow structure. The occurrence of particle-particle collisions becomes also more and more important with the increase of the local void fraction of the particulate phase. With increase of the particle loading the interaction with the fluid phase can not been neglected and 2-way or even 4-way coupling between the continous and disperse phases has to be taken into account. For dilute to moderate dense particle flows the Euler-Lagrange method is capable to resolve the main flow mechanism. An accurate computation needs unfortunately a high number of numerical particles (1,...,10^7) to get the reliable statistics for the underlying modelling correlations. Due to the fact that a Lagrangian algorithm cannot be vectorized for complex meshes the only way to finish those simulations in a reasonable time is the parallization applying the message passing paradigma. Frank et al. describes the basic ideas for a parallel Eulererian-Lagrangian solver, which uses multigrid for acceleration of the flow equations. The performance figures are quite good, though only steady problems are tackled. The presented paper is aimed to the numerical prediction of time-dependend fluid-particle flows using the simultanous particle tracking approach based on the Eulerian-Lagrangian and the particle-source-in-cell (PSI-Cell) approach. It is shown in the paper that for the unsteady flow prediction efficiency and load balancing of the parallel numerical simulation is an even more pronounced problem in comparison with the steady flow calculations, because the time steps for the time integration along one particle trajectory are very small per one time step of fluid flow integration and so the floating point workload on a single processor node is usualy rather low. Much time is spent for communication and waiting time of the processors, because for cold flow particle convection not very extensive calculations are necessary. One remedy might be a highspeed switch like Myrinet or Dolphin PCI/SCI (500 MByte/s), which could balance the relative high floating point performance of INTEL PIII processors and the weak capacity of the Fast-Ethernet communication network (100 Mbit/s) of the Chemnitz Linux Cluster (CLIC) used for the presented calculations. Corresponding to the discussed examples calculation times and parallel performance will be presented. Another point is the communication of many small packages, which should be summed up to bigger messages, because each message requires a startup time independently of its size. Summarising the potential of such a parallel algorithm, it will be shown that a Beowulf-type cluster computer is a highly competitve alternative to the classical main frame computer for the investigated Eulerian-Lagrangian simultanous particle tracking approach.

Ce mémoire présente le travail de recherche effectué pour la conception d'une stratégie de commande originale, destinée aux systèmes de puissance hybrides en sites isolés. Le système considéré, voué à l'alimentation électrique d'une habitation, comprend deux sources, un groupe de panneaux photovoltaïques et une petite éolienne, et deux types de stockage, un banc de batteries lithium-ion et un de supercondensateurs. Face au problème de gestion de l'énergie dans un système hybride, et aux enjeux de maximisation de sa puissance produite, nous proposons de développer une stratégie de commande basée sur les flux d'énergie. pour cela, nous présentons dans un premier temps les modélisations d'Euler-Lagrange et hamiltonienne du système. Ces modèles permettent d'utiliser la propriété de passivité de celui-ci, et ainsi de synthétiser des commandes par injection d'amortissement pour chaque source, afin de maximiser sa production, et pour les supercondensateurs, dans le but d'assurer une répartition cohérente des flux d'énergie entre eux et les batteries. Les commandes sont finalement mises en œuvre dans un simulateur, puis dans un banc d'essai expérimental, afin d'une part de comparer leurs performances à celles de solutions préexistantes, et d'autre part de valider le bon fonctionnement du système hybride complet les utilisant. / This thesis presents the research about design of a new control strategy for stand-alone hybrid power systems. The considered system is composed of two sources, a group of photovoltaic panels and a low-power wind generator, and of two kinds of storage, a bank of lithium-ion batteries and one of supercapacitors. Faced with the problem of energy management in a hybrid power system, and with necessity of maximizing the produced power, we intend to develop an energy-based control strategy.For this purpose, we present the system's Euler-Lagrange modeling and Hamiltonian modeling. These models allow the use of the passivity property, and then the design of Passivity-Based Controllers for each source, in order to maximize its production, and for the supercapacitors, to ensure a fitted power sharing between batteries and them. The controllers are finally implemented in a simulator, and then in a experimental test bench, in order to compare their performances to pre-existent solutions, and tovalidate the control law for the global hybrid system.

L'objectif de cette thèse est de contribuer à l'optimisation de problèmes dynamiques en présence de retard. Le point de vue qui nous intéressera est celui de Pontryagin qui dans son ouvrage publié en 1962 a donné les conditions nécessaires d'existence de solutions pour ce type de problème. Warga dans son ouvrage publié en 1972 a fait un catalogue des solutions possible, Li et al. ont étudié le cas de contrôle périodique. Notre méthode de démonstration est directement inspirée de la démonstration de P. Michel du cas des systèmes gouvernés par des équations différentielles ordinaires. La principale difficulté pour cette approche est l'utilisation de la résolvante de l'équation différentielle fonctionnelle linéarisée de l'équation différentielle fonctionnelle d'évolution qui gouverne le système. Nous traitons aussi de condition d'Euler-Lagrange dans le cadre d'un problème de calcul variationnel avec retard. / In this thesis, we have attempted to contribute to the optimization of dynamical problems with delay in state space. We are specifically interested in the viewpoint of Pontryagin who outlined in his book published in 1962 the necessary conditions required for solving such problems. In his work published in 1972, Warga catalogued the possible solutions. Li and al. analyzed the case of periodic control. We will treat an optimal control problem governed by a Delay Functional Differential Equation. Our method is close to the one of P. Michel on dynamical system governed by Ordinary Differential Equations. The main problem ariving out in this approach is the use of the resolvent of the Delay Functional Differential Equation. We also consider with Euler-Lagrange condition in the framework of variational problems with delay.

In power plants using large utility coal-fired boilers for generation of electricity the coal is pulverised in coal mills and then it has to be pneumatically transported and distributed to a larger number of burners (e.g. 30-40) circumferentially arranged in several rows around the burning chamber of the boiler. Besides the large pipework flow splitting devices are necessary for distribution of an equal amount of pulverised fuel (PF) to each of the burners. So called trifurcators (without inner fittings or guiding vanes) and ''riffle'' type bifurcators are commonly used to split the gas-coal particle flow into two or three pipes/channels with an equal amount of PF mass flow rate in each outflow cross section of the flow splitting device. These PF flow splitting devices are subject of a number of problems. First of all an uneven distribution of PF over the burners of a large utility boiler leads to operational and maintenance problems, increased level of unburned carbon and higher rates of NOX emissions. Maldistribution of fuel between burners caused by non uniform concentration of the PF (particle roping) in pipe and channel bends prior to flow splitting devices leads to uncontrolled differences in the fuel to air ratio between burners. This results in localised regions in the furnace which are fuel rich, where insufficient air causes incomplete combustion of the fuel. Other regions in the furnace become fuel lean, forming high local concentrations of NOX due to the high local concentrations of O2. Otherwise PF maldistribution can impact on power plant maintenance in terms of uneven wear on PF pipework, flow splitters as well as the effects on boiler panels (PF deposition, corrosion, slagging). In order to address these problems in establishing uniform PF distribution over the outlet cross sections of flow splitting devices in the pipework of coal-fired power plants the present paper deals with numerical prediction and analysis of the complex gas and coal particle (PF) flow through trifurcators and ''riffle'' type bifurcators. The numerical investigation is based on a 3-dimensional Eulerian- Lagrangian approach (MISTRAL/PartFlow-3D) developed by Frank et al. The numerical method is capable to predict isothermal, incompressible, steady gas- particle flows in 3-dimensional, geometrically complex flow geometries using boundary fitted, block-structured, numerical grids. Due to the very high numerical effort of the investigated gas-particle flows the numerical approach has been developed with special emphasis on efficient parallel computing on clusters of workstations or other high performance computing architectures. Besides the aerodynamically interaction between the carrier fluid phase and the PF particles the gas-particle flow is mainly influenced by particle-wall interactions with the outer wall boundaries and the inner fittings and guiding vanes of the investigated flow splitting devices. In order to allow accurate quantitative prediction of the motion of the disperse phase the numerical model requires detailed information about the particle-wall collision process. In commonly used physical models of the particle-wall interaction this is the knowledge or experimental prediction of the restitution coefficients (dynamic friction coefficient, coefficient of restitution) for the used combination of particle and wall material, e.g. PF particles on steel. In the present investigation these parameters of the particle-wall interaction model have been obtained from special experiments in two test facilities. Basic experiments to clarify the details of the particle-wall interaction process were made in a test facility with a spherical disk accelerator. This test facility furthermore provides the opportunity to investigate the bouncing process under normal pressure as well as under vacuum conditions, thus excluding aerodynamically influences on the motion of small particles in the near vicinity of solid wall surfaces (especially under small angles of attack). In this experiments spherical glass beads were used as particle material. In a second test facility we have investigated the real impact of non-spherical pulverised fuel particles on a steel/ceramic target. In this experiments PF particles were accelerated by an injector using inert gas like e.g. CO2 or N2 as the carrier phase in order to avoid dust explosion hazards. The obtained data for the particle-wall collision models were compared to those obtained for glass spheres, where bouncing models are proofed to be valid. Furthermore the second test facility was used to obtain particle erosion rates for PF particles on steel targets as a function of impact angles and velocities. The results of experimental investigations has been incorporated into the numerical model. Hereafter the numerical approach MISTRAL/PartFlow-3D has been applied to the PF flow through a ''riffle'' type bifurcator. Using ICEM/CFD-Hexa as grid generator a numerical mesh with approximately 4 million grid cells has been designed for approximation of the complex geometry of the flow splitting device with all its interior fittings and guiding vanes. Based on a predicted gas flow field a large number of PF particles are tracked throughout the flow geometry of the flow-splitter. Besides mean quantities of the particle flow field like e.g. local particle concentrations, mean particle velocities, distribution of mean particle diameter, etc. it is now possible to obtain information about particle erosion on riffle plates and guiding vanes of the flow splitting device. Furthermore the influence of different roping patterns in front of the flow splitter on the uniformness of PF mass flow rate splitting after the bifurcator has been investigated numerically. Results show the efficient operation of the investigated bifurcator in absence of particle roping, this means under conditions of an uniform PF particle concentration distribution in the inflow cross section of the bifurcator. If particle roping occurs and particle concentration differs over the pipe cross section in front of the bifurcator the equal PF particle mass flow rate splitting can be strongly deteriorated in dependence on the location and intensity of the particle rope or particle concentration irregularities. The presented results show the importance of further development of efficient rope splitting devices for applications in coal-fired power plants. Numerical analysis can be used as an efficient tool for their investigation and further optimisation under various operating and flow conditions.

The transport or the separation of solid particles or droplets suspended in a fluid flow is a common task in mechanical and process engineering. To improve machinery and physical processes (e.g. for coal combustion, reduction of NO_x and soot) an optimization of complex phenomena by simulation applying the fundamental conservation equations is required. Fluid-particle flows are characterized by the ratio of density of the two phases gamma=rho_P/rho_F, by the Stokes number St=tau_P/tau_F and by the loading in terms of void and mass fraction. Those numbers (Stokes number, gamma) define the flow regime and which relevant forces are acting on the particle. Dependent on the geometrical configuration the particle-wall interaction might have a heavy impact on the mean flow structure. The occurrence of particle-particle collisions becomes also more and more important with the increase of the local void fraction of the particulate phase. With increase of the particle loading the interaction with the fluid phase can not been neglected and 2-way or even 4-way coupling between the continous and disperse phases has to be taken into account. For dilute to moderate dense particle flows the Euler-Lagrange method is capable to resolve the main flow mechanism. An accurate computation needs unfortunately a high number of numerical particles (1,...,10^7) to get the reliable statistics for the underlying modelling correlations. Due to the fact that a Lagrangian algorithm cannot be vectorized for complex meshes the only way to finish those simulations in a reasonable time is the parallization applying the message passing paradigma. Frank et al. describes the basic ideas for a parallel Eulererian-Lagrangian solver, which uses multigrid for acceleration of the flow equations. The performance figures are quite good, though only steady problems are tackled. The presented paper is aimed to the numerical prediction of time-dependend fluid-particle flows using the simultanous particle tracking approach based on the Eulerian-Lagrangian and the particle-source-in-cell (PSI-Cell) approach. It is shown in the paper that for the unsteady flow prediction efficiency and load balancing of the parallel numerical simulation is an even more pronounced problem in comparison with the steady flow calculations, because the time steps for the time integration along one particle trajectory are very small per one time step of fluid flow integration and so the floating point workload on a single processor node is usualy rather low. Much time is spent for communication and waiting time of the processors, because for cold flow particle convection not very extensive calculations are necessary. One remedy might be a highspeed switch like Myrinet or Dolphin PCI/SCI (500 MByte/s), which could balance the relative high floating point performance of INTEL PIII processors and the weak capacity of the Fast-Ethernet communication network (100 Mbit/s) of the Chemnitz Linux Cluster (CLIC) used for the presented calculations. Corresponding to the discussed examples calculation times and parallel performance will be presented. Another point is the communication of many small packages, which should be summed up to bigger messages, because each message requires a startup time independently of its size. Summarising the potential of such a parallel algorithm, it will be shown that a Beowulf-type cluster computer is a highly competitve alternative to the classical main frame computer for the investigated Eulerian-Lagrangian simultanous particle tracking approach.

Many fluid flow processes in nature and technology are characterized by the presence and coexistence of two ore more phases. These two- or multiphase flows are furthermore characterized by a greater complexity of possible flow phenomena and phase interactions then in single phase flows and therefore the numerical simulation of these multiphase flows is usually demanding a much higher numerical effort. The presented work summarizes the research and development work of the author and his research group on "Numerical Methods for Multiphase Flows" at the University of Technology, Chemnitz over the last years. This work was focussed on the development and application of numerical approaches for the prediction of disperse fluid-particle flows in the field of fluid mechanics and process technology. A main part of the work presented here is concerned with the modelling of different physical phenomena in fluid-particle flows under the paradigm of the Lagrangian treatment of the particle motion in the fluid. The Eulerian-Lagrangian approach has proved to be an especially well suited numerical approach for the simulation of disperse multiphase flows. On the other hand its application requires a large amount of (parallel) computational power and other computational ressources. The models described in this work give a mathematical description of the relevant forces and momentum acting on a single spherical particle in the fluid flow field, the particle-wall interaction and the particle erosion to the wall. Further models has been derived in order to take into account the influence of particle-particle collisions on the particle motion as well as the interaction of the fluid flow turbulence with the particle motion. For all these models the state-of-the-art from literature is comprehensively discussed. The main field of interest of the work presented here is in the area of development, implementation, investigation and comparative evaluation of parallelization methods for the Eulerian-Lagrangian approach for the simulation of disperse multiphase flows. Most of the priorly existing work of other authors is based on shared-memory approaches, quasi-serial or static domain decomposition approaches. These parallelization methods are mostly limited in theire applicability and scalability to parallel computer architectures with a limited degree of parallelism (a few number of very powerfull compute nodes) and to more or less homogeneous multiphase flows with uniform particle concentration distribution and minor complexity of phase interactions. This work now presents a novel parallelization method developed by the author, realizing a dynamic load balancing for the Lagrangian approach (DDD - Dynamic Domain Decomposition) and therefore leading to a substantial decrease in total computation time necessary for multiphase flow computations with the Eulerian-Lagrangian approach. Finally, the developed and entirely parallelized Eulerian-Lagrangian approach MISTRAL/PartFlow-3D offers the opportunity of efficient investigation of disperse multiphase flows with higher concentrations of the disperse phase and the resulting strong phase interaction phenomena (four-way coupling). / Viele der in Natur und Technik ablaufenden Strömungsvorgänge sind durch die Koexistenz zweier oder mehrerer Phasen gekennzeichnet. Diese sogenannten Zwei- oder Mehrphasensysteme zeichnen sich durch ein hohes Maß an Komplexität aus und erfordern oft einen sehr hohen rechentechnischen Aufwand zu deren numerischer Simulation. Die vorliegende Arbeit faßt langjährige Forschungs- und Entwicklungsarbeiten des Autors und seiner Forschungsgruppe "Numerische Methoden für Mehrphasenströmungen" an der TU Chemnitz zusammen, die sich mit der Entwicklung und Anwendung numerischer Berechnungsverfahren für disperse Fluid-Partikel-Strömungen auf dem Gebiet der Strömungs- und Verfahrenstechnik befassen. Ein wesentlicher Teil der Arbeit befaßt sich mit der Modellierung unterschiedlicher physikalischer Phänomene in Fluid-Partikel-Strömungen unter dem Paradigma der Lagrange'schen Betrachtungsweise der Partikelbewegung. Das Euler-Lagrange-Verfahren hat sich als besonders geeignetes Berechnungsverfahren für die numerische Simulation disperser Mehrphasenströmungen erwiesen, stellt jedoch in seiner Anwendung auch höchste Anforderungen an die Ressourcen der verwendeten (parallelen) Rechnerarchitekturen. Die näher ausgeführten mathematisch-physikalischen Modelle liefern eine Beschreibung der auf eine kugelförmige Einzelpartikel im Strömungsfeld wirkenden Kräfte und Momente, der Partikel-Wand-Wechselwirkung und der Partikelerosion. Weitere Teilmodelle dienen der Berücksichtigung von Partikel-Partikel-Stoßvorgängen und der Wechselwirkung zwischen Fluidturbulenz und Partikelbewegung. Der Schwerpunkt dieser Arbeit liegt im Weiteren in der Entwicklung, Untersuchung und vergleichenden Bewertung von Parallelisierungsverfahren für das Euler-Lagrange-Verfahren zur Berechnung von dispersen Mehrphasenströmungen. Zuvor von anderen Autoren entwickelte Parallelisierungsmethoden für das Lagrange'sche Berechnungsverfahren basieren im Wesentlichen auf Shared-Memory-Ansätzen, Quasi-Seriellen Verfahren oder statischer Gebietszerlegung (SDD) und sind somit in ihrer Einsetzbarkeit und Skalierbarkeit auf Rechnerarchitekturen mit relativ geringer Parallelität und auf weitgehend homogene Mehrphasenströmungen mit geringer Komplexität der Phasenwechselwirkungen beschränkt. In dieser Arbeit wird eine vom Autor entwickelte, neuartige Parallelisierungsmethode vorgestellt, die eine dynamische Lastverteilung für das Lagrange-Verfahren ermöglicht (DDD - Dynamic Domain Decomposition) und mit deren Hilfe eine deutliche Reduzierung der Gesamtausführungszeiten einer Mehrphasenströmungsberechnung mit dem Euler-Lagrange-Verfahren möglich ist. Im Ergebnis steht mit dem vom Autor und seiner Forschungsgruppe entwickelten vollständig parallelisierten Euler-Lagrange-Verfahren MISTRAL/PartFlow-3D ein numerisches Berechnungsverfahren zur Verfügung, mit dem disperse Mehrphasenströmungen mit höheren Konzentrationen der dispersen Phase und daraus resultierenden starken Phasenwechselwirkungen (Vier-Wege-Kopplung) effektiv untersucht werden können.