This thesis is focused on the issue of computational modeling of soil harvesting while using the compact excavator from the company DOOSAN BOBCAT EMEA s.r.o.. The solution consists in creating two computational models in Rocky DEM and Ansys Mechanical. Rocky DEM software is used to solve the disconnection of soil by excavator components using the discrete element method. The outcome includes courses of forces and moments during the particular time of the individual joints of the model that was used. The obtained courses are then applied as an external load to the joint of a selected structural part of the analyzed model using a kinematic model in ANSYS Mechanical. The kinematic model consists of simplified geometry models of individual parts of the excavator arm, and a more detailed geometry model of the analyzed part of the arm. All the parts are connected to each other by rotational bonds representing joints. A static structural analysis of the mechanical stress is performed in ANSYS Mechanical for the prepared model during the simulated process. Finally, the selected structural part is assessed with respect to the elastic limit and fatigue strength.

The interaction between fluids and structures, which is an interdisciplinary problem, has gained importance in a wide range of scientific and engineering applications. Thanks to new advances in computer technology, the numerical analysis of multiphysics phenomena has aroused growing interest. Fluid-structure interactions have been numerically and experimentally studied by many researchers and published by several books, papers, and review papers. Hou et al. (2012) [3] have also published a review paper entitled “Numerical methods for fluid-structure interaction”, which provides useful knowledge about different approaches for FSI analysis. The key challenge encountered in any numerical FSI analysis is the coupling between the two independent domains with clear distinctions. For example, a structure domain requires discretizing by a Lagrangian mesh where the mesh is fixed to the mass and follows the mass motion. In fact, the Lagrangian mesh is able to deform and follows an individual structural mass as it moves through space and time. Nonetheless, the fluid mesh remains intact within the space, where the fluid flows as time passes. The numerical approaches with regard to FSI phenomena can be divided into two main categories, namely the monolithic approach and the partitioned approach. In the former, a single system equation for the whole problem is solved simultaneously by a unified algorithm; however, in the latter, the fluid and the structure are discretized with their proper mesh and solved separately by different numerical algorithms. When a fluid flow interacts with a structure, the pressure load arising from the fluid flow is exerted on the structure, followed by deformations, stresses, and strains of the structure. Depending on the resulting deformation and the rate of the variations, a one-way or two-way coupling analysis can be conducted. Fluid-structure interaction (FSI) is characterized by the interaction of some movable or deformable structure with an internal or surrounding fluid flow. In a fluid-structure interaction (FSI), the laws that describe fluid dynamics and structural mechanics are coupled. There is also another classification for FSI problems on the basis of mesh methods: conforming methods and non-conforming methods. In the first method, the interface condition is regarded as a physical boundary (interface boundary) moving during the solution time, which imposes the mesh for the fluid domain to be updated in conformity with the new position for the interface. In contrast, the implementation of the second method eliminates a need for the fluid mesh update on the account of the fact that the interface requirement is enforced by constraints on the system equations instead of the physical boundary motion. In this work, we study numerically and experimentally the fluid-structure interaction comprising a flexible slender shaped structure, free surface flow and potentially interacting rigid structures, categorized in flood protection applications, whereas more emphasis is given to numerical analysis. Objectives of this study are defined in detail as follows: The initial aim is the numerical analysis of the behavior of a down-scale membrane loaded by hydrostatic pressures, where the numerical results have to be validated against available experimental data. A further case which has to be investigated is how the full scale flexible flood barrier behaves when approached and impacted by an accelerated massive flotsam. The numerical model has to be built so as to replicate the same physical phenomenon investigated experimentally. It enables a comparison between the numerical and experimental analyses to be drawn. A more complicated case where the flexible down-scale membrane interacts with a propagated water wave is a further target area to study. Moreover, an experimental investigation is required to validate the numerical results by way of comparison. The ultimate goal is to perform a similitude analysis upon which a correlation between the full-scale prototype and the down-scale model can be formed. The implementation of the similarity laws enables the behavior of the full scale prototype to be quantitatively assessed on the basis of the available data for the down-scale model. In addition, in order to validate the accuracy of the similitude analysis, numerical analyses have to be carried out.:Contents Zusammenfassung I ABSTRACT IV Nomenclature X 1 Introduction 1 1.1 Work overview 2 1.2 Literature review 3 1.2.1 The non-conforming methods 6 1.2.2 The conforming (partitioned) approaches 11 1.2.2.1 Interface data transfer 16 1.2.2.2 Accuracy, stability and efficiency 16 1.2.2.3 Modification of interface conditions: Robin transmission conditions 18 1.3 Concluding remarks 19 2 Methodology-numerical methods for fluid-structure interaction analysis (FSI) 20 2.1 Single FV framework 21 2.1.1 The prism layer mesher 24 2.1.2 Turbulence modeling 24 2.2 Preparation of the standalone Abaqus model 27 2.2.1 Damping by bulk viscosity 28 2.2.2 Coulomb friction damping 29 2.2.3 Rayleigh damping 29 2.2.4 Determination of the Rayleigh damping parameters based on the Chowdhury procedure 29 2.2.5 The frequency response function (FRF) measurement 30 2.2.6 The half-power bandwidth method 31 2.3 Explicit partitioned coupling 33 2.4 Implicit partitioned coupling 39 2.5 Overset mesh 40 2.6 Concluding remarks 42 3 Verification and validation of the structural model 44 3.1 Numerical model setup of the down-scale membrane 44 3.2 Comparing similarity between numerical and experimental results 46 3.2.1 Hypothesis test terminology 46 3.2.2 Curve fitting 47 3.2.3 Similarity measures between two curves 48 3.3 Results (down-scale membrane) 52 3.3.1 Similarity tests for the contact length 54 3.3.2 Similarity tests for the slope 58 3.3.3 Similarity tests for the displacement in Y direction 60 3.4 Concluding remarks 63 4 Numerical model setup of the original membrane for impact analysis 66 4.1 Structure domain 67 4.2 Fluid domain 72 4.2.1 Standard mesh and results 74 4.2.2 Overset mesh 80 4.3 Co-simulation model setup and results 88 4.4 Concluding remarks 96 5 Numerical wave generation 100 5.1 Theoretical estimation of the waves 107 5.2 Numerical wave tank setup 110 5.3 Results 114 5.4 Concluding remarks 119 6 Validity of the model with dynamic pressure 121 6.1 Wave tank 123 6.2 Structure domain 127 6.3 Fluid domain 130 6.4 Co-simulation model setup 136 6.5 Experimental approach 137 6.6 Results 141 6.6.1 Similarity tests for the displacement of the membrane in X direction 156 6.6.2 Similarity tests for the displacement of the membrane in Y direction 160 6.6.3 Similarity tests for the displacement of the membrane in Z direction 164 6.7 Concluding remarks 168 7 Similarity 171 7.1 Motivation 171 7.2 Governing equations 174 7.3 Buckingham Pi theorem 175 7.4 Dimensionless numbers 175 Similitude requirement 177 7.5 Simulation setup 178 7.6 Results 179 7.7 Concluding remarks 191 8 Summary, conclusions and outlook 192 List of figures 199 List of tables 209 References 210

Die vorliegende Arbeit beschäftigt sich mit Festkörperkontinuumsmechanik und Metall- bzw. Kristallplastizität auf verschiedenen Längenskalen. Diesbezüglich besteht die Arbeit aus drei größeren Teilen. Im ersten Teil werden Verformungsvorgänge mit expliziter FEM (Finite-Elemente-Methode) und einem makroskopischen phänomenologischen Modell der Viskoplastizität simuliert. Hierbei wird sich auf das Gleichkanalwinkelpressen (ECAP) eines Metallbarrens und die Stauchung einer sogenannten Crashbox konzentriert. In beiden Fällen gelingt es, die im Experiment bereits beobachtete Lokalisierung der Verformung korrekt wiederzugeben. Da bei den Simulationen die konkrete Mikrostruktur des Materials vernachlässigt wird, werden diese Lokalisierungsphänomene als verformungsinduziert angesehen. Der zweite Teil beschäftigt sich mit der Erweiterung des viskoplastischen Modells, sodass mikroskopische Vorgänge der Gitterdefektstruktur des Materials berücksichtigt werden können. Dazu wird ein Modell des dynamischen Verhaltens von Versetzungspopulationen entwickelt und an das makroskopische viskoplastische Modell gekoppelt. Auf diese Weise können Aspekte der sogenannten Kornfeinung – einem komplexen Strukturbildungsprozess von Versetzungen und anderen Gitterdefekten – erfasst werden. Allerdings kann die für die makroskopischen Eigenschaften entscheidende Bildung von Subkorngrenzen auf diese Weise nicht abgebildet werden. Um dies zu erreichen, wird im dritten Teil der Arbeit eine mesoskopische Theorie der Kristallplastizität mit kontinuierlich verteilten Versetzungen verwendet und weiterentwickelt. Hierbei werden die für eine Subkornbildung wesentlichen Freiheitsgrade hinzugenommen, die Anzahl phänomenologischer Ansätze und zugehöriger Materialparameter aber so klein wie möglich gehalten. Mit dieser Kontinuumsversetzungstheorie (KVT) gelingt es, die Bildung von Subkorngrenzen bei großen plastischen Verformungen eines Kristallits zu verfolgen. Bei den impliziten FEM-Simulationen wird ebenfalls eine Lokalisierung beobachtet, allerdings in Bezug auf die Aktivität der Versetzungen in verschiedenen Gleitebenen. Dementsprechend wird dieses Lokalisierungsphänomen als versetzungsinduziert angesehen. Der Beitrag der vorliegenden Arbeit liegt zum einen in der Aufarbeitung und Gegenüberstellung unterschiedlicher methodischer Herangehensweisen zur Modellierung verformungs- und versetzungsinduzierter Lokalisierungsphänomene. Zum anderen wird eine Analyse und Vereinheitlichung der geometrisch linearen KVT nach Berdichevsky & Le vorgenommen. Wie sich dabei zeigt, verhindern inhärente kinematische Einschränkungen der Theorie die Simulation einer Subkornbildung. Aus diesem Grund wird die konsistente geometrisch nichtlineare KVT von Gurtin aufgegriffen und erweitert. Mit einem daraus abgeleiteten elastisch und plastisch anisotropen Modell der Einkristallviskoplastizität wird der Nachweis erbracht, dass die Subkornbildung damit simuliert werden kann. Darüber hinaus wird eine Aufbereitung und Synthese von Algorithmen zur numerischen Lösung der zugehörigen Feldgleichungen mittels der Methode der finiten Differenzen und der finiten Elemente geliefert. Zudem werden beide Näherungsverfahren in Bezug auf Vor- und Nachteile sowie thermodynamische Konsistenz bei der Anwendung auf Mehrfeldprobleme miteinander verglichen. / The present thesis deals with solid continuum mechanics applied to metal and crystal plasticity on different length scales. In this respect, the work consists of three larger parts. In the first part, deformation processes are simulated with explicit FEM (Finite Element Method) and a macroscopic phenomenological model of viscoplasticity. Here the focus is on the Equal-Channel Angular Pressing (ECAP) of a metal billet and the compression of a so-called crash box. In both cases it is possible to correctly reproduce the localization of the deformation as already observed in the experiment. Since the concrete microstructure of the material is neglected in the simulations, these localization phenomena are regarded as deformation-induced. The second part deals with the extension of the viscoplastic model so that microscopic processes of the lattice defect structure of the material can be considered. A model of the dynamic behavior of dislocation populations is developed and coupled to the macroscopic viscoplastic model. In this way, aspects of the so-called grain refinement – a complex structure formation process of dislocations and other lattice defects – can be captured. However, the formation of subgrain boundaries, which is decisive for the macroscopic properties, cannot be predicted in this way. To achieve this, a mesoscopic theory of crystal plasticity with continuously distributed dislocations is used and further developed in the third part of the thesis. Here, the degrees of freedom essential for subgrain formation are added, while the number of phenomenological approaches and associated material parameters are kept as small as possible. With this continuum dislocation theory it is possible to follow the formation of subgrain boundaries during large plastic deformations of a crystallite. In the implicit FEM simulations, localization is also observed, but with respect to the dislocation activity in different slip planes. Accordingly, this localization phenomenon is considered dislocation-induced. The contribution of the present work lies on the one hand in the review and comparison of different methodical approaches to the modeling of deformation- and dislocation-induced localization phenomena. On the other hand, an analysis and unification of the geometrically linear continuum dislocation theory according to Berdichevsky & Le is carried out. As it turns out, inherent kinematic limitations of the theory prevent the simulation of subgrain formation. For this reason the consistent geometrically non-linear continuum dislocation theory from Gurtin is adopted and extended. With the derived model of elastically and plastically anisotropic single crystal viscoplasticity it is proven that subgrain formation can be simulated. Moreover, a preparation and synthesis of algorithms for the numerical solution of the associated field equations using the method of finite differences and finite elements is provided. In addition, both approximation methods are compared in terms of advantages and disadvantages as well as thermodynamic consistency when applied to multi-field problems.