Dinamica di una imbarcazione da canottaggio: simulazioni numeriche con un modello RANS
Tuesday 22nd July 2008
This Master Thesis is part of a research project concerning the development of a mathematical and numerical model for the dynamics of olimpic rowing sculls in the works at the Laboratory for Modeling and Scientific Computing MOX of Politecnico of Milan in collaboration with Filippi Lido S.r.l.. In recent years (see (FMMP08) and (FMMM08)) a complete model for dynamics of rowing sculls has been developed. The model requires the definition of: • The athlets movement during rowing cycle; • The force at the oars; • The hydrodynamic forces on the scull. On one hand a rather exact formulation for rowers movement has been optained from motion capture data and a consistent profile for oars forces has been measured; on the other hand, concerning the hydrodynamics forces, the original formulation contains several approximations. In fact, for the sake of efficiency, the original model extimates drag forces by standard formulas based on drag-coefficient of scull and wetted surface and approximates lift and torque forces through the hydrostatic configuration of the scull. The dissipative effects of waves generated by the secondary movements are dealt with by imposing a damping term obtained by the solution of a potential flow problem. In this work a method for coupling the dynamic model for rowing scull with a RANSE (Reynolds Averaged Navier Stokes Equation) solver is studied. Based on this approach it is possible to determine hydrodynamic forces and moments accurately by solving the flow surrounding the scull. This objective is reached by the implementation of a FSI (Fluid Structure Interaction) method which combines the commercial state of the art CFD solver Fluent with the code for rowing dynamics Kime developed at MOX. The study of a series of intermediate problems has been necessary to finally achieve the fluid-structure interaction method. At first, a computational grid has been realized. The study of viscous turbulent freesurface flows around a hull requires a high resolution grid in the boundary layer of the hull and in the interface zone between air and water. On this grid, a steady state computation has been performed. Then, a method for moving the hull in the computational domain has been studied using suitable techniques of dynamic mesh. A method for the imposition of the horizontal acceleration due to a time-dependent rowing propulsion has also been studied. These methods have been widely analyzed for elementary motion in the different degrees of freedom of the boat. In the next phase the dynamics of a rowing boat was gathered from a Kime standalone (no FSI) simulation and imposed to the hull in the computational domain. By the imposition of motion one can verify that mesh deformation does not affect the convergence of fluid computation. Finally, the coupling between the fluid solver and the rigid body dynamical system was completed, realizing in this way a model for the dynamics of rowing boat where hydrodynamcs forces are computed by a RANSE finite volume solver. The innovative part of this work can be evaluated in two components: on the one hand we complete the model for dynamic of rowing scull with a computation of fluid forces whose accuracy rely only on grid resolution; on the other hand we simulate in a RANSE solver the six degrees of freedom dynamics of a time-dependent propulsion boat. The disadvantage of such a method consists in very long computational time necessary to simulation: on a rather coarse grid we need 6 hours on a single AMD Opteron 64 bit 1.8GHz processor to perform 1 second of simulation, while the stand-alone Kime version takes only few minutes to perform 1 minute of simulation. This problem makes the FSI method more useful for calibrate the Kime code in hydrodynamics forces modeling section rather than a really usable software for shape and crew configuration optimisation. Actually the implementation of a high efficient parallel version of the code is being investigated in order to perform simulations with more computing resources and reduce computational time. The outline of the work is the following. The first chapter gives an overview of the problem and supplies some references about existing literature on the dynamics of rowing boat and on boat dynamics simulation with RANSE finite volume codes. Chapter 2 exposes the mathematical model describing the motion of viscous turbulent free-surface flows and in Chapter 3 we show the discretization of such equations based on a finite volume approach. Chapter 4 introduces the model for the dynamics of rowing boat and the proposed coupling with the fluid dynamics solver. In Chapter 5 we show the requirement for a good mesh generation in our problem and the proposed strategy to generate the hull movement in the computational domain. Finally, in Chapter 6 we present the numerical results obtained in all the phases of our work while a paragraph with some conclusions and suggestion for continuing the work concludes this thesis.