The natural soils and sedimentary rocks are typically formed by deposition and progressive consolidation of marine sediments. Consequently, they are characterized by the presence of closely spaced bedding planes, resulting in anisotropy in their mechanical behaviour. Among anisotropic rocks, the group of sedimentary rocks known as shales is of a particular interest as it is often the host rock in nuclear waste storage and oil industry. The Tournemire shales are anisotropic in terms of deformability and the failure mode, which means that complex constitutive models should be used to describe their mechanical response. In this thesis a pragmatic methodology based on the notion of a microstructure tensor, as suggested by Pietruszczak and Mroz (2001), has been employed for the description of orientation dependent characteristics of Tournemire shale. This has been combined with a plasticity framework that incorporates an anisotropic deviatoric hardening. The formulation requires identification of several parameters including strength descriptors associated with the failure criterion and constants that are involved in describing the anisotropy and strain hardening. All the material functions/parameters have been identified here based on the experimental results reported by Niandou et al. (1997). Using those parameters, the numerical simulations of a number of triaxial tests were conducted and the results compared with the experimental data in order to verify the performance of the model. After the verification stage, the formulation was incorporated in a commercial FE code (Abaqus/standard) using the UMAT interface and was then applied to a numerical analysis of a tunnel excavation within the anisotropic rock mass. The numerical results, including the distribution of the damage and vertical/horizontal displacements, have been compared for different orientations of the bedding planes. / Thesis / Master of Applied Science (MASc)

This thesis is a study of massive star formation: the environments in which they form and the effect that their radiation feedback has on their environments. We present high-performance supercomputer simulations of massive star formation inside molecular cloud clumps and cores. First, we present a novel radiative transfer code that hybridizes two previous approaches to radiative transfer (raytracing and flux-limited diffusion) and implements it in a Cartesian grid-based code with adaptive mesh refinement, representing the first of such implementations. This hybrid radiative transfer code allows for more accurate calculations of the radiation pressure and irradiated gas temperature that are the hallmark of massive star formation and which threaten to limit the mass which stars can ultimately obtain. Next, we apply this hybrid radiative transfer code in simulations of massive protostellar cores. We simulate their gravitational collapse and the formation of a massive protostar surrounded by a Keplerian accretion disk. These disks become gravitationally unstable, increasing the accretion rate onto the star, but do not fragment to form additional stars. We demonstrate that massive stars accrete material predominantly through their circumstellar disks, and via radiation pressure drive large outflow bubbles that appear stable to classic fluid instabilities. Finally, we present simulations of the larger context of star formation: turbulent, magnetised, filamentary cloud clumps. We study the magnetic field geometry and accretion flows. We find that in clouds where the turbulent and magnetic energies are approximately equal, the gravitational energy must dominate the kinetic energy for there to be a coherent magnetic field structure. Star cluster formation takes place inside the primary filament and the photoionisation feedback from a single massive star drives the creation of a bubble of hot, ionised gas that ultimately engulfs the star cluster and destroys the filament. / Thesis / Doctor of Philosophy (PhD)

This thesis presents a series of direct numerical simulations (DNS) performed in order to understand the discrepancy in the literature regarding turbulent asymptotic suction boundary layers (TASBLs) at low Reynolds numbers. The hypothesis to be tested is that the main reason for higher turbulence intensities observed in experiments compared to DNS is that the latter exhibits intermittent patches of laminar flow, developing both temporally and spatially. This hypothesis is confirmed here by comparing simulations with and without tripping, where the former removed patches of laminar flow thereby establishing a fully developed turbulent state with higher turbulence intensities compared to its naturally developing counterpart. The DNS were performed at different suctions rates corresponding to Reynolds numbers above the critical value of 270. The statistics taken from the simulations at different streamwise positions also support the developing character of the flow with increasing intermittency further downstream. Thus, it can be concluded that the actual flow state at these marginal Reynolds numbers is indeed an intermittent one, with lower fluctuation values as the experimental data would indicate.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.