Graduation date: 1999

This dissertation details the development of a block model for the movement of submarine slides with emphasis on possible hydroplaning. Unlike previous models, the block model simulated the mechanism of hydroplaning by monitoring the contact condition between the bottom surface of the slide mass and the underlying ground. The effect of hydroplaning on the movement of the slide mass is considered by changing the forces applied on the slide mass by the underlying ground according to the contact condition. The hydrodynamic stresses applied on the slide mass by the surrounding fluid are determined based on the numerical simulations of the flow around a sliding mass. The sliding process of the block is disretisized in a step-by-step manner using a Newmark scheme. A computer program is also written to implement the block model. The block model is validated by comparisons between the numerical results and data reported by Mohrig, et al (1999) for laboratory experiments on subaqueous slides. An illustrative study is also conducted using the block model for the movement of the sediment slabs during the Storegga Slide. The block model has successfully predicted the occurrence of hydroplaning and run-out distances of subaqueous slides. Numerical results with the block model supports the mechanism of hydroplaning for subaqueous slides with greater run-out distances than comparable subaerial slides.

Not available / text

The increasing number of discovered extra-solar planets opens a new opportunity for studies of the formation of planetary systems. Their diversity keeps challenging the long-standing theories which were based on data primarily from our own solar system. Resonant planetary systems are of particular interest because their dynamical configuration provides constraints on the otherwise unobservable formation and migration phase. In this thesis, formation scenarios for the planetary systems HD128311 and HD45364 are presented. N-body simulations of two planets and two dimensional hydrodynamical simulations of proto-planetary discs are used to realistically model the convergent migration phase and the capture into resonance. The results indicate that the proto-planetary disc initially has a larger surface density than previously thought. Proto-planets are exposed to stochastic forces, generated by density fluctuations in a turbulent disc. A generic model of both a single planet, and two planets in mean motion resonance, being stochastically forced is presented and applied to the system GJ876. It turns out that GJ876 is stable for reasonable strengths of the stochastic forces, but systems with lighter planets can get disrupted. Even if a resonance is not completely disrupted, stochastic forces create characteristic, observable libration patterns. As a further application, the stochastic migration of small bodies in Saturn’s rings is studied. Analytic predictions of collisional and gravitational interactions of a moonlet with ring particles are compared to realistic three dimensional collisional N-body simulations with up to a million particles. Estimates of both the migration rate and the eccentricity evolution of embedded moonlets are confirmed. The random walk of the moonlet is fast enough to be directly observable by the Cassini spacecraft. Turbulence in the proto-stellar disc also plays an important role during the early phases of the planet formation process. In the core accretion model, small, metre-sized particles are getting concentrated in pressure maxima and will eventually undergo a rapid gravitational collapse to form a gravitationally bound planetesimal. Due to the large separation of scales, this process is very hard to model numerically. A scaled method is presented, that allows for the correct treatment of self-gravity for a marginally collisional system by taking into account the relevant small scale processes. Interestingly, this system is dynamically very similar to Saturn’s rings.

Many free surface flow problems encountered in hydraulic engineering can be accurately analyzed by utilizing the depth-averaged equations of motion. A consequence of adopting this depth-averaged modeling approach is that closure approximations must be implemented to represent the so-called effective stresses. These effective stresses consist of the depth-averaged viscous stresses, which are usually small and therefore neglected, the depth-averaged turbulent Reynold's stresses, and additional stresses resulting from depth-averaging of the nonlinear 'convective acceleration terms (often called momentum dispersion terms). Attention is focused on examining closure for both the depth-averaged Reynold's stresses and the momentum dispersion terms. In the present study, the penalty function finite element technique is utilized to solve the governing hydrodynamic and turbulence model equations for a variety of flow domains. Alternative momentum dispersion and turbulence closure models are proposed and evaluated by comparing model predictions with experimental data for strongly curved open channel flow. The results of these simulations indicate that the depth-averaged (k-ε) turbulence model yields excellent agreement with experimental observations. In addition, it appears that neither the streamline curvature modification of the depth-averaged (k-ε) model, nor the momentum dispersion models based on the assumption of helicoidal flow in a curved channel, yield significant improvement in model predictions. Overall model predictions are found to be as good as those of a more complex and restricted three dimensional model. / Ph. D.

La non-reproductibilité numérique apparait dans divers domaines d'application de la simulation HPC. En effet, les différentes distributions d'un calcul parallèle peuvent mener à des résultats numériques différents, à cause des particularités de l'arithmétique flottante. Le besoin de reproductibilité numérique est motivé pour le débogage, le test et la validation des codes de calcul scientifique. Nous nous intéressons aux simulations par éléments finis en hydrodynamique implémentées dans le logiciel openTelemac qui est largement utilisé pour des applications industrielles et scientifiques. Nous identifions et analysons les sources de cette non-reproductibilité. Nous définissons et implementons comment récupérer la reproductibilité numérique de deux modules d'openTelemac. Nous mesurons que le sur-coût en terme de temps de calcul de la version reproductible est tout à fait raisonnable en pratique. / HPC simulations in various scientific domains suffer from failures of numerical reproducibility because of floating-point arithmetic peculiarities. Different distributions of a parallel computation may yield different numerical results. Numerical reproducibility is a requested feature to facilitate the debug, the validation and the test of industrial or large software. In this thesis, we focus on the openTelemac software that implements finite element simulation for industrial and scientific hydrodynamics. We identify and analyze the sources of this reproducibility failure. We define and implement how to recover numerical reproducibility in two openTelemac modules. We also measure that the running time extra-cost of the reproducible version is reasonable enough in practice.