Discontinuities can appear in different fields of mechanics. Some examples where discontinuities arise are more obvious such as the formation of cracks. Other sources of discontinuities are less apparent such as interfaces between different materials. Furthermore continuous fields with steep gradients can also be considered as discontinuous fields. This work aims at the inclusion of arbitrary discontinuities within the finite element method. Although the finite element method is the most sophisticated numerical tool in modern engineering, the inclusion of discontinuities is still a challenging task. Traditionally within finite the framework of FE methods discontinuities are modeled explicitely by the construction of the mesh. Thus, when a fixed mesh is used, the position of the discontinuity is prescribed by the location of interelement boundaries and not by the physical situation. The simulation of crack growth requires a frequent adaption of the mesh and that can be a difficult and computationally expensive task. Thus a more flexible numerical approach is needed which leads to the mesh-independent representation of the discontinuity. A challenging field where the accurate description of discontinuities is of vital importance is the modeling of failure in engineering materials. The load capacity of a structure is limited by the material strength. If the load limit is exceeded failure zones arise and increase. Representative examples of failure mechanisms are are cracks in brittle materials or shear bands in metals or soils. Failure processes are often accompanied by a strain softening material behaviour (decreasing load carrying capacity with increasing strain at a material point). It is known that the inclusion of strain softening material behaviour within a continuum description requires regularization techniques to preserve the well- posedness of the governing equations. One possibility is the consideration of non-local or gradient terms in the constitutive equations but these approaches require a sufficiently fine discretization in the localization zone, which leads to a high numerical effort. If the extent of the failure zone and the failure process to the point of the development of discrete cracks is considered it seems reasonable to include strong discontinuities. In the framework of fracture mechanics the inclusion of displacement jumps is intuitively comprehensible. However, the modeling of localized failure processes demands the consideration of inelastic material behaviour. Cohesive zone models represent an approach which is especially suited for the incorporation within the finite element framework. It is supposed that cohesive tractions are transmitted between the discontinuity surfaces. These tractions are constitutively prescribed by a phenomenological traction separation law and thus allow for the modeling of different inelastic mechanisms, like micro-crack evolution, initiation of voids, plastic flow or crack bridging. The incorporation of a displacement discontinuity in combination with a cohesive traction separation law leads to a sound model to describe failure processes and crack propagation. Another area where the existence of discontinuities is not as obvious is the occurence of material interfaces, inclusions or holes. The accurate modeling of such internal interfaces is important to predict the mechanical behaviour of components. The present discontinuity is of different nature: the displacement field is continuous but there is a jump in the strains, which is denoted by the expression weak discontinuity. Usually in FE methods material interfaces are taken into account by the mesh construction. But if the structure exhibits multiple inclusions of complex geometry it can be advantageous if the interface does not have to be meshed. And when we look at at problems where the interface moves with time, e. g. phase transformation, the mesh-independent modeling of the weak discontinuities naturally holds major advantages. The greatest challenge in the modeling of discontinuities is their incorporation into numerical methods. The focus of the present work is the development, analysis and application of a finite element approach to model mesh-independent discontinuities. The method shall be robust and flexible to be applicable to both, strong and weak discontinuities.
In contrast to the spatial motion setting, the material motion setting of continuum mechanics is concerned with the response to variations of material placements of particles with respect to the ambient material. The material motion point of view is thus extremely prominent when dealing with defect mechanics to which it has originally been introduced by Eshelby more than half a century ago. Its primary unknown, the material deformation map is governed by the material motion balance of momentum, i.e. the balance of material forces on the material manifold in the sense of Eshelby. Material (configurational) forces are concerned with the response to variations of material placements of 'physical particles' with respect to the ambient material. Opposed to that, the common spatial (mechanical) forces in the sense of Newton are considered as the response to variations of spatial placements of 'physical particles' with respect to the ambient space. Material forces as advocated by Maugin are especially suited for the assessment of general defects as inhomogeneities, interfaces, dislocations and cracks, where the material forces are directly related to the classical J-Integral in fracture mechanics, see also Gross & Seelig. Another classical example of a material - or rather configurational - force is emblematized by the celebrated Peach-Koehler force, see e.g. the discussion in Steinmann. The present work is mainly divided in four parts. In the first part we will introduce the basic notions of the mechanics and numerics of material forces for a quasi-static conservative mechanical system. In this case the internal potential energy density per unit volume characterizes a hyperelastic material behaviour. In the first numerical example we discuss the reliability of the material force method to calculate the vectorial J-integral of a crack in a Ramberg-Osgood type material under mode I loading and superimposed T-stresses. Secondly, we study the direction of the single material force acting as the driving force of a kinked crack in a geometrically nonlinear hyperelastic Neo-Hooke material. In the second part we focus on material forces in the case of geometrically nonlinear thermo-hyperelastic material behaviour. Therefore we adapt the theory and numerics to a transient coupled problem, and elaborate the format of the Eshelby stress tensor as well as the internal material volume forces induced by the gradient of the temperature field. We study numerically the material forces in a bimaterial bar under tension load and the time dependent evolution of material forces in a cracked specimen. The third part discusses the material force method in the case of geometrically nonlinear isotropic continuum damage. The basic equations are similar to those of the thermo-hyperelastic problem but we introduce an alternative numerical scheme, namely an active set search algorithm, to calculate the damage field as an additional degree of freedom. With this at hand, it is an easy task to obtain the gradient of the damage field which induces the internal material volume forces. Numeric examples in this part are a specimen with an elliptic hole with different semi-axis, a center cracked specimen and a cracked disc under pure mode I loading. In the fourth part of this work we elaborate the format of the Eshelby stress tensor and the internal material volume forces for geometrically nonlinear multiplicative elasto-plasticity. Concerning the numerical implementation we restrict ourselves to the case of geometrically linear single slip crystal plasticity and compare here two different numerical methods to calculate the gradient of the internal variable which enters the format of the internal material volume forces. The two numerical methods are firstly, a node point based approach, where the internal variable is addressed as an additional degree of freedom, and secondly, a standard approach where the internal variable is only available at the integration points level. Here a least square projection scheme is enforced to calculate the necessary gradients of this internal variable. As numerical examples we discuss a specimen with an elliptic inclusion and an elliptic hole respectively and, in addition, a crack under pure mode I loading in a material with different slip angles. Here we focus on the comparison of the two different methods to calculate the gradient of the internal variable. As a second class of numerical problems we elaborate and implement a geometrically linear von Mises plasticity with isotropic hardening. Here the necessary gradients of the internal variables are calculated by the already mentioned projection scheme. The results of a crack in a material with different hardening behaviour under various additional T-stresses are given.
This thesis aims at an overall improvement of the diffusion coefficient predictions. For this reason the theoretical determination of diffusion, viscosity, and thermodynamics in liquid systems is discussed. Furthermore, the experimental determination of diffusion coefficients is also part of this work. All investigations presented are carried out for organic binary liquid mixtures. Diffusion coefficient data of 9 highly nonideal binary mixtures are reported over the whole concentration range at various temperatures, (25, 30, and 35) °C. All mixtures investigated in a Taylor dispersion apparatus consist of an alcohol (ethanol, 1-propanol, or 1-butanol) dissolved in hexane, cyclohexane, carbon tetrachloride, or toluene. The uncertainty of the reported data is estimated to be within 310-11 m2s-1. To compute the thermodynamic correction factor an excess Gibbs energy model is required. Therefore, the applicability of COSMOSPACE to binary VLE predictions is thoroughly investigated. For this purpose a new method is developed to determine the required molecular parameters such as segment types, areas, volumes, and interaction parameters. So-called sigma profiles form the basis of this approach which describe the screening charge densities appearing on a molecule’s surface. To improve the prediction results a constrained two-parameter fitting strategy is also developed. These approaches are crucial to guarantee the physical significance of the segment parameters. Finally, the prediction quality of this approach is compared to the findings of the Wilson model, UNIQUAC, and the a priori predictive method COSMO-RS for a broad range of thermodynamic situations. The results show that COSMOSPACE yields results of similar quality compared to the Wilson model, while both perform much better than UNIQUAC and COSMO-RS. Since viscosity influences also the diffusion process, a new mixture viscosity model has been developed on the basis of Eyring’s absolute reaction rate theory. The nonidealities of the mixture are accounted for with the thermodynamically consistent COSMOSPACE approach. The required model and component parameters are derived from sigma-profiles, which form the basis of the a priori predictive method COSMO-RS. To improve the model performance two segment parameters are determined from a least-squares analysis to experimental viscosity data, whereas a constraint optimisation procedure is applied. In this way the parameters retain their physical meaning. Finally, the viscosity calculations of this approach are compared to the findings of the Eyring-UNIQUAC model for a broad range of chemical mixtures. These results show that the new Eyring-COSMOSPACE approach is superior to the frequently employed Eyring-UNIQUAC method. Finally, on the basis of Eyring’s absolute reaction rate theory a new model for the Maxwell-Stefan diffusivity has been developed. This model, an extension of the Vignes equation, describes the concentration dependence of the diffusion coefficient in terms of the diffusivities at infinite dilution and an additional excess Gibbs energy contribution. This energy part allows the explicit consideration of thermodynamic nonidealities within the modelling of this transport property. If the same set of interaction parameters, which has been derived from VLE data, is applied for this part and for the thermodynamic correction, a theoretically sound modelling of VLE and diffusion can be achieved. The influence of viscosity and thermodynamics on the model accuracy is thoroughly investigated. For this purpose diffusivities of 85 binary mixtures consisting of alkanes, cycloalkanes, halogenated alkanes, aromatics, ketones, and alcohols are computed. The average relative deviation between experimental data and computed values is approximately 8 % depending on the choice of the gE-model. These results indicate that this model is superior to some widely used methods. In summary, it can be said that the new approach facilitates the prediction of diffusion coefficients. The final equation is mathematically simple, universally applicable, and the prediction quality is as good as other models recently developed without having to worry about additional parameters, like pure component physical property data, self diffusion coefficients, or mixture viscosities. In contrast to many other models, the influence of the mixture viscosity can be omitted. Though a viscosity model is not required in the prediction of diffusion coefficients with the new equation, the models presented in this work allow a consistent modelling approach of diffusion, viscosity, and thermodynamics in liquid systems.
Within the last decades, a remarkable development in materials science took place -- nowadays, materials are not only constructed for the use of inert structures but rather designed for certain predefined functions. This innovation was accompanied with the appearance of smart materials with reliable recognition, discrimination and capability of action as well as reaction. Even though ferroelectric materials serve smartly in real applications, they also possess several restrictions at high performance usage. The behavior of these materials is almost linear under the action of low electric fields or low mechanical stresses, but exhibits strong non-linear response under high electric fields or mechanical stresses. High electromechanical loading conditions result in a change of the spontaneous polarization direction with respect to individual domains, which is commonly referred to as domain switching. The aim of the present work is to develop a three-dimensional coupled finite element model, to study the rate-independent and rate-dependent behavior of piezoelectric materials including domain switching based on a micromechanical approach. The proposed model is first elaborated within a two-dimensional finite element setting for piezoelectric materials. Subsequently, the developed two-dimensional model is extended to the three-dimensional case. This work starts with developing a micromechanical model for ferroelectric materials. Ferroelectric materials exhibit ferroelectric domain switching, which refers to the reorientation of domains and occurs under purely electrical loading. For the simulation, a bulk piezoceramic material is considered and each grain is represented by one finite element. In reality, the grains in the bulk ceramics material are randomly oriented. This property is taken into account by applying random orientation as well as uniform distribution for individual elements. Poly-crystalline ferroelectric materials at un-poled virgin state can consequently be characterized by randomly oriented polarization vectors. Energy reduction of individual domains is adopted as a criterion for the initiation of domain switching processes. The macroscopic response of the bulk material is predicted by classical volume-averaging techniques. In general, domain switching does not only depend on external loads but also on neighboring grains, which is commonly denoted as the grain boundary effect. These effects are incorporated into the developed framework via a phenomenologically motivated probabilistic approach by relating the actual energy level to a critical energy level. Subsequently, the order of the chosen polynomial function is optimized so that simulations nicely match measured data. A rate-dependent polarization framework is proposed, which is applied to cyclic electrical loading at various frequencies. The reduction in free energy of a grain is used as a criterion for the onset of the domain switching processes. Nucleation in new grains and propagation of the domain walls during domain switching is modeled by a linear kinetics theory. The simulated results show that for increasing loading frequency the macroscopic coercive field is also increasing and the remanent polarization increases at lower loading amplitudes. The second part of this work is focused on ferroelastic domain switching, which refers to the reorientation of domains under purely mechanical loading. Under sufficiently high mechanical loading, however, the strain directions within single domains reorient with respect to the applied loading direction. The reduction in free energy of a grain is used as a criterion for the domain switching process. The macroscopic response of the bulk material is computed for the hysteresis curve (stress vs strain) whereby uni-axial and quasi-static loading conditions are applied on the bulk material specimen. Grain boundary effects are addressed by incorporating the developed probabilistic approach into this framework and the order of the polynomial function is optimized so that simulations match measured data. Rate dependent domain switching effects are captured for various frequencies and mechanical loading amplitudes by means of the developed volume fraction concept which relates the particular time interval to the switching portion. The final part of this work deals with ferroelectric and ferroelastic domain switching and refers to the reorientation of domains under coupled electromechanical loading. If this free energy for combined electromechanical loading exceeds the critical energy barrier elements are allowed to switch. Firstly, hysteresis and butterfly curves under purely electrical loading are discussed. Secondly, additional mechanical loads in axial and lateral directions are applied to the specimen. The simulated results show that an increasing compressive stress results in enlarged domain switching ranges and that the hysteresis and butterfly curves flatten at higher mechanical loading levels.
This thesis deals with the development of thermoplastic polyolefin elastomers using recycled polyolefins and ground tyre rubber (GTR). The disposal of worn tyres and their economic recycling mean a great challenge nowadays. Material recycling is a preferred way in Europa owing to legislative actions and ecological arguments. This first step with worn tyres is already done in this direc-tion as GTR is available in different fractions in guaranteed quality. As the traditional applications of GTR are saturated, there is a great demand for new, value-added products containing GTR. So, the objective of this work was to convert GTR by reac-tive blending with polyolefins into thermoplastic elastomers (TPE) of suitable me-chanical and rheological properties. It has been established that bituminous reclamation of GTR prior to extrusion melt compounding with polyolefins is a promising way of TPE production. By this way the sol-content (acetone soluble fraction) of the GTR increases and the GTR particles can be better incorporated in the corresponding polyolefin matrix. The adhesion be-tween GTR and matrix is given by molecular intermingling in the resulting interphase. GTR particles of various production and mean particle size were involved in this study. As polyolefins recycled low-density polyethylene (LDPE), recycled high-density polyethylene (HDPE) and polypropylene (PP) were selected. First, the opti-mum conditions for the GTR reclamation in bitumen were established (160 °C < T < 180 °C; time ca. 4 hours). Polyolefin based TPEs were produced after GTR reclamation in extrusion compounding. Their mechanical (tensile behaviour, set properties), thermal (dynamic-mechanical thermal analysis, differential scanning calorimetry) and rheological properties (both in low- and high-shear rates ) were determined. The PE-based blends contained an ethylene/propylene/diene (EPDM) rubber as compatibilizer and their composition was as follows: PE/EPDM/GTR:bitumen = 50/25/25:25. The selected TPEs met the most important criterion, i.e. elongation at break > 100 %; compression set < 50%. The LDPE-based TPE (TPE(LDPE)) showed better me-chanical performance compared to the TPE(HDPE). This was assigned to the higher crystallinity of the HDPE. The PP-based blends of the compositions PP/(GTR-bitumen) 50/50 and 25/75, whereby the ratio of GTR/bitumen was 60/40, outperformed those containing non-reclaimed GTR. The related blends showed also a better compatibility with a PP-based commercial thermoplastic dynamic vulcanizate (TDV). Surprisingly, the mean particle size of the GTR, varied between < 0.2 and 0.4-0.7 mm, had a small effect on the mechanical properties, however somewhat larger for the rheological behaviour of the TPEs produced.
In the present work, various aspects of the mixed continuum-atomistic modelling of materials are studied, most of which are related to the problems arising due to a development of microstructures during the transition from an elastic to plastic description within the framework of continuum-atomistics. By virtue of the so-called Cauchy-Born hypothesis, which is an essential part of the continuum-atomistics, a localization criterion has been derived in terms of the loss of infinitesimal rank-one convexity of the strain energy density. According to this criterion, a numerical yield condition has been computed for two different interatomic energy functions. Therewith, the range of the Cauchy-Born rule validity has been defined, since the strain energy density remains quasiconvex only within the computed yield surface. To provide a possibility to continue the simulation of material response after the loss of quasiconvexity, a relaxation procedure proposed by Tadmor et al. leading necessarily to the development of microstructures has been used. Thereby, various notions of convexity have been overviewed in details. Alternatively to the above mentioned criterion, a stability criterion has been applied to detect the critical deformation. For the study in the postcritical region, the path-change procedure proposed by Wagner and Wriggers has been adapted for the continuum-atomistic and modified. To capture the deformation inhomogeneity arising due to the relaxation, the Cauchy-Born hypothesis has been extended by assumption that it represents only the 1st term in the Taylor's series expansion of the deformation map. The introduction of the 2nd, quadratic term results in the higher-order materials theory. Based on a simple computational example, the relevance of this theory in the postcritical region has been shown. For all simulations including the finite element examples, the development tool MATLAB 6.5 has been used.
The polydispersive nature of the turbulent droplet swarm in agitated liquid-liquid contacting equipment makes its mathematical modelling and the solution methodologies a rather sophisticated process. This polydispersion could be modelled as a population of droplets randomly distributed with respect to some internal properties at a specific location in space using the population balance equation as a mathematical tool. However, the analytical solution of such a mathematical model is hardly to obtain except for particular idealized cases, and hence numerical solutions are resorted to in general. This is due to the inherent nonlinearities in the convective and diffusive terms as well as the appearance of many integrals in the source term. In this work two conservative discretization methodologies for both internal (droplet state) and external (spatial) coordinates are extended and efficiently implemented to solve the population balance equation (PBE) describing the hydrodynamics of liquid-liquid contacting equipment. The internal coordinate conservative discretization techniques of Kumar and Ramkrishna (1996a, b) originally developed for the solution of PBE in simple batch systems are extended to continuous flow systems and validated against analytical solutions as well as published experimental droplet interaction functions and hydrodynamic data. In addition to these methodologies, we presented a conservative discretization approach for droplet breakage in batch and continuous flow systems, where it is found to have identical convergence characteristics when compared to the method of Kumar and Ramkrishna (1996a). Apart from the specific discretization schemes, the numerical solution of droplet population balance equations by discretization is known to suffer from inherent finite domain errors (FDE). Two approaches that minimize the total FDE during the solution of the discrete PBEs using an approximate optimal moving (for batch) and fixed (for continuous systems) grids are introduced (Attarakih, Bart & Faqir, 2003a). As a result, significant improvements are achieved in predicting the number densities, zero and first moments of the population. For spatially distributed populations (such as extraction columns) the resulting system of partial differential equations is spatially discretized in conservative form using a simplified first order upwind scheme as well as first and second order nonoscillatory central differencing schemes (Kurganov & Tadmor, 2000). This spatial discretization avoids the characteristic decomposition of the convective flux based on the approximate Riemann Solvers and the operator splitting technique required by classical upwind schemes (Karlsen et al., 2001). The time variable is discretized using an implicit strongly stable approach that is formulated by careful lagging of the nonlinear parts of the convective and source terms. The present algorithms are tested against analytical solutions of the simplified PBE through many case studies. In all these case studies the discrete models converges successfully to the available analytical solutions and to solutions on relatively fine grids when the analytical solution is not available. This is accomplished by deriving five analytical solutions of the PBE in continuous stirred tank and liquid-liquid extraction column for especial cases of breakage and coalescence functions. As an especial case, these algorithms are implemented via a windows computer code called LLECMOD (Liquid-Liquid Extraction Column Module) to simulate the hydrodynamics of general liquid-liquid extraction columns (LLEC). The user input dialog makes the LLECMOD a user-friendly program that enables the user to select grids, column dimensions, flow rates, velocity models, simulation parameters, dispersed and continuous phases chemical components, and droplet phase space-time solvers. The graphical output within the windows environment adds to the program a distinctive feature and makes it very easy to examine and interpret the results very quickly. Moreover, the dynamic model of the dispersed phase is carefully treated to correctly predict the oscillatory behavior of the LLEC hold up. In this context, a continuous velocity model corresponding to the manipulation of the inlet continuous flow rate through the control of the dispersed phase level is derived to get rid of this behavior.
A general framework for the thermodynamics of open systems is developed in the spatial and the material setting. Special emphasis is placed on the balance of mass which is enhanced by additional source and flux terms. Different solution strategies within the finite element technique are derived and compared. A number of numerical examples illustrates the features of the proposed approach.
In this thesis, the enhanced Galerkin (eG) finite element method in time is presented. The eG method leads to higher order accurate energy and momentum conserving time integrators for the underlying finite-dimensional Hamiltonian systems. This thesis is concerned with particle dynamics and semi-discrete nonlinear elastodynamics. The conservation is generally related to the collocation property of the eG method. The momentum conservation renders the Gaussian quadrature and the energy conservation is obtained by using a new projection technique. An objective time discretisation of the used strain measures avoids artificial strains for large superimposed rigid body motions. The numerical examples show the well long term performance in the presence of stiffness as well as for calculating large-strain motions.
The goal of this thesis is a physically motivated and thermodynamically consistent formulation of higher gradient inelastic material behavior. Thereby, the influence of the material microstructure is incorporated. Next to theoretical aspects, the thesis is complemented with the algorithmic treatment and numerical implementation of the derived model. Hereby, two major inelastic effects will be addressed: on the one hand elasto-plastic processes and on the other hand damage mechanisms, which will both be modeled within a continuum mechanics framework.