| The research work in this dissertation focuses on Smoothed Particle Hydrodynamics (SPH), a fully Lagrangian particle method for modelling the behavior of fluids in a computational framework. The main goal is to use and improve aspects of this Computational Fluid Dynamics (CFD) technique for the solution of novel problems of hydrodynamics including free surface flows. To this extent, simulations of several CFD problems are presented and the accuracy of the SPH results is compared with numerous works in the cited literature.;A SPH investigation of the wave and bottom pressure fields generated by a fast hull in finite-depth water is conducted for different flow conditions. This subject is relevant to several engineering studies, therefore an assessment of the correlation between water waves and pressure disturbances at the sea floor is made in a quantitative manner. Furthermore, SPH is used for the first time to study harmonic oscillations of a thin rigid lamina in a viscous fluid with and without a free surface. The results provide useful insights about the hydrodynamic load dependence on phenomena of vortex shedding and advection, which are well captured by the SPH method. Finally, a functional relationship is introduced to express the total fluid force as added mass and damping coefficients and to quantify their dependence on the control parameters governing the problem.;Computational findings are also used to identify the shortcomings of SPH and part of the research is devoted to understanding and improving these aspects. Particularly, two current SPH challenges are of interest in this dissertation: the first is concerned with the concept of adaptivity in SPH and the testing of particle refinement/de-refinement techniques for introducing higher resolutions in areas where detailed flow information is critical (e.g. boundary layer, free surface, stagnation areas). To this extent, numerical solutions of a dam break and sloshing problems are presented, corroborating an increase in the accuracy of the simulations and a substantial gain in computational time with respect to using uniformly distributed particles. The second major direction of improvement regards the development of a new type of boundary condition to simulate open boundaries in SPH. This is not a trivial task in Lagrangian methods as particles must be inserted in/removed from the domain while preserving physical quantities and numerical conditions such as stability and consistency. Simulations presented herein are validated against several other works in the literature for both 2-D and 3-D problems, highlighting a good agreement among the results.;Overall, data from the computational studies in this research supports the importance of both adaptivity and open boundary conditions towards achieving accurate SPH solutions of real-life engineering problems. |
