Numerical modelling of fluid-structure interaction with application in hemodynamics

The thesis deals with numerical algorithms for fluid-structure interaction problems with application in blood flow modelling. It starts with a short introduction on the mathematical description of incompressible viscous flow with non-Newtonian viscosity and a moving linear viscoelastic structure. The mathematical model consists of the generalized Navier-Stokes equation used for the description of fluid flow and the generalized string model for structure movement. The arbitrary Lagrangian-Eulerian approach is used in order to take into account moving computational domain. A part of the thesis is devoted to the discussion on the non-Newtonian behaviour of shear-thinning fluids, which is in our case blood, and derivation of two non-Newtonian models frequently used in the blood flow modelling. Further we give a brief overview on recent fluid-structure interaction schemes with discussion about the difficulties arising in numerical modelling of blood flow. Our main contribution lies in numerical and experimental study of a new loosely-coupled partitioned scheme called the kinematic splitting fluid-structure interaction algorithm. We present stability analysis for a coupled problem of non-Newtonian shear-dependent fluids in moving domains with viscoelastic boundaries. Here, we assume both, the nonlinearity in convective as well is diffusive term. We analyse the convergence of proposed numerical scheme for a simplified fluid model of the Oseen type. Moreover, we present series of experiments including numerical error analysis, comparison of hemodynamic parameters for the Newtonian and non-Newtonian fluids and comparison of several physiologically relevant computational geometries in terms of wall displacement and wall shear stress. Numerical analysis and extensive experimental study for several standard geometries confirm reliability and accuracy of the proposed kinematic splitting scheme in order to approximate fluid-structure interaction problems.

Application of reactive fluid transport modeling to hydrothermal systems

A one-dimensional multi-component reactive fluid transport algorithm, 1DREACT (Steefel, 1993) was used to investigate different fluid-rock interaction systems. A major short coming of mass transport calculations which include mineral reactions is that solid solutions occurring in many minerals are not treated adequately. Since many thermodynamic models of solid solutions are highly non-linear, this can seriously impact on the stability and efficiency of the solution algorithms used. Phase petrology community saw itself faced with a similar predicament 10 years ago. To improve performance and reliability, phase equilibrium calculations have been using pseudo compounds. The same approach is used here in the first, using the complex plagioclase solid solution as an example. Thermodynamic properties of a varying number of intermediate plagioclase phases were calculated using ideal molecular, Al-avoidance, and non-ideal mixing models. These different mixing models can easily be incorporated into the simulations without modification of the transport code. Simulation results show that as few as nine intermediate compositions are sufficient to characterize the diffusional profile between albite and anorthite. Hence this approach is very efficient, and can be used with little effort. A subsequent chapter reports the results of reactive fluid transport modeling designed to constrain the hydrothermal alteration of Paleoproterozoic sediments of the Southern Lake Superior region. Field observations reveal that quartz-pyrophyllite (or kaolinite) bearing assemblages have been transformed into muscovite-pyrophyllite-diaspore bearing assemblages due to action of fluids migrating along permeable flow channels. Fluid-rock interaction modeling with an initial qtz-prl assemblage and a K-rich fluid simulates the formation of observed mineralogical transformation. The bulk composition of the system evolves from an SiO2-rich one to an Al2O3+K2O-rich one. Simulations show that the fluid flow was up-temperature (e.g. recharge) and that fluid was K-rich. Pseudo compound approach to include solid solutions in reactive transport models was tested in modeling hydrothermal alteration of Icelandic basalts. Solid solutions of chlorites, amphiboles and plagioclase were included as the secondary mineral phases. Saline and fresh water compositions of geothermal fluids were used to investigate the effect of salinity on alteration. Fluid-rock interaction simulations produce the observed mineral transformations. They show that roughly the same alteration minerals are formed due to reactions with both types of fluid which is in agreement with the field observations. A final application is directed towards the remediation of nitrate rich groundwaters. Removal of excess nitrate from groundwater by pyrite oxidation was modeled using the reactive fluid transport algorithm. Model results show that, when a pyrite-bearing, permeable zone is placed in the flow path, nitrate concentration in infiltrating water can be significantly lowered, in agreement with proposals from the literature. This is due to nitrogen reduction. Several simulations investigate the efficiency of systems with different mineral reactive surface areas, reactive barrier zone widths, and flow rates to identify the optimum setup.

Synthesis and characterization of structurally well-defined polymer-layered silicate nanocomposites

Zusammenfassung Nanokomposite aus Polymeren und Schichtsilikaten werden zumeist auf der Basis natürlicher Tone wie Montmorillonit hergestellt. Für NMR- und EPR-Untersuchungen der Tensidschicht, die das Silikat mit dem Polymer kompatibilisiert, ist der Eisengehalt natürlicher Tone jedoch abträglich, weil er zu einer Verkürzung der Relaxationszeiten und zu einer Linienverbreiterung in den Spektren führt. Dieses Problem konnte überwunden werden, indem als Silikatkomponente eisenfreies, strukturell wohldefiniertes Magadiit hydrothermal synthetisiert und für die Kompositbildung eingesetzt wurde. Die Morphologie des Magadiits wurde durch Rasterelektronenmikroskopie charakterisiert und der Interkalationsgrad von schmelzinterkalierten Polymer-Nanokompositen wurde durch Weitwinkelröntgenstreuung bestimmt. Polymere mit Carbonylgruppen scheinen leichter zu interkalieren als solche ohne Carbonylgruppen. Polycaprolacton interkalierte sowohl in Oragnomagadiite auf der Basis von Ammoniumtensiden als auch in solche auf der Basis von Phosphoniumtensiden. Die Dynamik auf einer Nanosekundenzeitskala und die Struktur der Tensidschicht wurden mittels ortsspezifisch spinmarkierter Tensidsonden unter Nutzung von Dauerstrich- (CW) und Puls-Methoden der elektronenparamagnetischen Resonanzspektroskopie (EPR) untersucht. Zusätzlich wurde die statische 2H-Kernmagnetresonanz (NMR) an spezifisch deuterierten Tensiden angewendet, um die Tensiddynamik auf einer komplementären Zeitskala zwischen Mikrosekunden und Millisekunden zu erfassen. Sowohl die CW-EPR- als auch die 2H-NMR-Ergebnisse zeigen eine Beschleunigung der Tensiddynamik durch Interkalation von Polycaprolacton auf, während sich in den nichtinterkalierten Mikrokompositen mit Polystyrol die Tensiddynamik verlangsamt. Die Rotationskorrelationszeiten und Aktivierungsenergien offenbaren verschiedene Regime der Tensiddynamik. In Polystyrol-Mikrokompositen entspricht die Übergangstemperatur zwischen den Regimen der Glasübergangstemperatur von Polystyrol, während sie in Polycaprolacton-Nanokompositen bei der Schmelztemperatur von Polycaprolacton liegt. Durch die erhebliche Verlängerung der Elektronenspin-Relaxationszeiten bei Verwendung von eisenfreiem Magadiit können Messdaten hoher Qualität mit Puls-EPR-Experimenten erhalten werden. Insebsondere wurden die Vier-Puls-Elektron-Elektron-Doppelresonanz (DEER), die Elektronenspinechoenveloppenmodulation (ESEEM) und die Elektronen-Kern-Doppelresonanz (ENDOR) an spinmarkierten sowie spezifisch deuterierten Tensiden angewandt. Die ENDOR-Ergebnisse legen ein Model der Tensidschicht nahe, in dem zusätzlich zu den Oberflächenlagen auf dem Silikat eine wohldefinierte mittlere Lage existiert. Dieses Modell erklärt auch Verdünnungseffekte durch das Polymer in Kompositen mit Polycaprolacton und Polystyrol. Die umfangreiche Information aus den Magnetresonanztechniken ergänzt die Information aus konventionellen Charakterisierungstechniken wie Röntgendiffraktion und Transmissionselektronenmikroskopie und führt so zu einem detaillierteren Bild der Struktur und Dynamik der Tensidschicht in Nanokompositen aus Polymeren und Schichtsilikaten.

Dynamic wetting of complex liquids

A simple dependency between contact angle θ and velocity or surface tension has been predicted for the wetting and dewetting behavior of simple liquids. According to the hydrodynamic theory, this dependency was described by Cox and Voinov as θ ∼ Ca^(1/3) (Ca: Capillary number). For more complex liquids like surfactant solutions, this prediction is not directly given.rnHere I present a rotating drum setup for studying wetting/dewetting processes of surfactant solutions on the basis of velocity-dependent contact angle measurements. With this new setup I showed that surfactant solutions do not follow the predicted Cox-Voinov relation, but showed a stronger contact angle dependency on surface tension. All surfactants independent of their charge showed this difference from the prediction so that electrostatic interactions as a reason could be excluded. Instead, I propose the formation of a surface tension gradient close to the three-phase contact line as the main reason for the strong contact angle decrease with increasing surfactant concentration. Surface tension gradients are not only formed locally close to the three-phase contact line, but also globally along the air-liquid interface due to the continuous creation/destruction of the interface by the drum moving out of/into the liquid. By systematically hindering the equilibration routes of the global gradient along the interface and/or through the bulk, I was able to show that the setup geometry is also important for the wetting/dewetting of surfactant solutions. Further, surface properties like roughness or chemical homogeneity of the wetted/dewetted substrate influence the wetting/dewetting behavior of the liquid, i. e. the three-phase contact line is differently pinned on rough/smooth or homogeneous/inhomogeneous surfaces. Altogether I showed that the wetting/dewetting of surfactant solutions did not depend on the surfactant type (anionic, cationic, or non-ionic) but on the surfactant concentration and strength, the setup geometry, and the surface properties.rnSurfactants do not only influence the wetting/dewetting behavior of liquids, but also the impact behavior of drops on free-standing films or solutions. In a further part of this work, I dealt with the stability of the air cushion between drop and film/solution. To allow coalescence between drop and substrate, the air cushion has to vanish. In the presence of surfactants, the vanishing of the air is slowed down due to a change in the boundary condition from slip to no-slip, i. e. coalescence is suppressed or slowed down in the presence of surfactant.

Non-aqueous dispersions of particles and their applications

In the current work, three studies about non-aqueous dispersions of particles were carried out by using an amphiphilic block copolymer poly(isoprene)-block-poly(methyl methacrylate) (PI-b-PMMA) as stabilizer:rn1. Dispersions of polyurethane and polyurea porous particles for polymer compositesrn2. Dispersions of PMMA and PU particles with PDI dye for study of Single Molecule Spectroscopy Detectionrn3. Dispersions of graphene nanosheets for polymer compositesrnrnIn the first study, highly porous polyurethane and polyurea particles were prepared in a non-aqueous emulsion. The preparation of porous particles consisted of two parts: At first, a system was developed where the emulsion had high stability for the polymerization among diisocyanate, diol and water. In the second part, porous particles were prepared by using two methods fission/fusion and combination by which highly porous particles were obtained. In this study, the applications of porous particles were also investigated where polyurethane particles were tested as filling material for polymer composites and as catalyst carrier for polyethylene polymerization. rnrnIn the second study, PMMA and PU particles from one non-aqueous emulsion were investigated via single molecule fluorescence detection. At first the particles were loaded with PDI dye, which were detected by fluorescence microscopy. The distribution and orientation of the PDI molecules in the particles were successfully observed by Single Molecule Fluorescence Detection. The molecules were homogenously distributed inside of the particles. In addition they had random orientation, meaning that no aggregations of dye molecules were formed. With the results, it could be supposed that the polymer chains were also homogenously distributed in the particles, and that the conformation was relatively flexible. rnrnIn the third part of the study, graphene nanosheets with high surface area were dispersed in an organic solvent with low boiling point and low toxicity, THF, stabilized with a block copolymer PI-b-PMMA. The dispersion was used to prepare polymer composites. It was shown that the modified graphene nanosheets had good compatibility with the PS and PMMA matrices. rn