Smoothed-particle-hydrodynamics simulation of port hydrodynamic problems

The thesis reports on the development of a numerical procedure based on the Lagrangian Smoothed-Particle-Hydrodynamics (SPH) method for the simulation of hydrodynamic problems in harbours. The target applications focus on ship induced scouring of the harbour bottom. Respective erosions represent unpleasant phenomena, especially if they occur close to quay walls. They can significantly weaken the structural support and cause cost intensive counter measures. These measures are usually based on a rather weak background knowledge, thus simulations might help to analyse the erosional processes without the need for model- or full-scale experiments. Contrary to common state-of-the-art simulation tools, the present work pursues a monolithic approach to capture the complex overall problem. This requires an SPH procedure that is able to cope with water/soil interaction, floating self-propelled ships as well as large computational domains. Within the present SPH-framework, water is modelled as a Newtonian fluid and the soil is treated by a combined solid/fluid-approach based on the Mohr-Coulomb yield criterion. The technique captures small deformations of the soil by an elastic model whereas a transition to a rate dependent non-Newtonian fluid behaviour is initiated for higher strain rates. The interaction between water and soil is realised by a three-layer suspension model which is needed for an accurate prediction of erosional phenomena. Partly saturated porous media can also play an important role in soil failure processes. They are taken into account by a saturation dependent variable cohesion with the seepage flow through the soil skeleton being evaluated from a simple Darcy approach. A bodyforce propulsor model based on open water characteristics is used to represent ship propellers and thrusters. Vessel motions are captured by a 6DOF motion solver. To enhance the code’s applicability to large domains, a variable particle resolution strategy based on changeable particle masses is applied. This allows for a fine resolution of local details whereas the far field can be modelled by a coarse particle distribution in order to reduce the overall number of particles.