Tesis:
Sloshing flows : experimental investigation and numerical simulations with smoothed particle hydrodynamics
- Autor: DELORME, Louis
- Título: Sloshing flows : experimental investigation and numerical simulations with smoothed particle hydrodynamics
- Fecha: 2008
- Materia: Sin materia definida
- Escuela: E.T.S. DE INGENIEROS NAVALES
- Departamentos: ARQUITECTURA Y CONSTRUCCION NAVALES
- Acceso electrónico: http://oa.upm.es/53701/
- Director/a 1º: SOUTO IGLESIAS, Antonio
- Resumen: Los fenómenos de sloshing, o “chapoteo”, se pueden definir como los movimientos de la superficie libre de un liquido contenido en un tanque. Generan cargas dinámicas sobre la estructura del tanque, cuya predicción es un gran desafió de los ingenieros trabajando sobre sloshing. Sus aplicaciones se encuentran en las ingenierías aeronáutica, nuclear, naval, y una literatura abundante sobre sloshing existe, recientemente resumida en el libro de Ibrahim (2005). En el mundo naval, los fenómenos de sloshing pueden tener efectos positivos o negativos. En el caso general, la presencia de un tanque con superficie libre disminuye la estabilidad estática del buque, sobre todo cuando se tratan de grandes cantidades de líquido, como es el caso de los metaneros, que pueden contener hasta más de 200000 m3 de gas líquido. El acoplamiento dinámico entre los movimientos del buque y aquellos del líquido dentro de los tanques es algo muy difícil de predecir, especialmente cuando se forman olas rompientes que impactan sobre la estructura de los tanques. Las presiones debidas a estos impactos son también difíciles de predecir, cuando el diseño de los tanques dependa de ellas. En las últimas décadas, numerosos estudios sobre modelos de tanques han sido realizados por los sociedades de clasificación como DNV (Berg, 1987),o ABS (Card & Hoseong, 2005). Los mayores problemas que se encuentran son el carácter aleatorio de las presiones de impacto y la dificultad de escalar los resultados entre el modelo y el prototipo a escala real (Bass et al., 1985). Por otra parte, los fenómenos de sloshing son buscados en el caso de tanques estabilizadores pasivos, que se montan sobre algunos barcos para disminuir su movimiento de balance. La idea de estos tanques se debe a Watts (1883) y consiste en que el líquido dentro del tanque crea un momento que se opone al momento debido a las olas, para aquellas con frecuencias vecinas de la de resonancia del buque en balance. Frahm (1911) introdujo los tanques con forma de U, reduciendo de este modo los efectos de superficie libre. En estos años, poca confianza existía en estos sistemas con grandes cantidades de líquido libre dentro del barco. En los últimos 60 años, con el creciente interés en los fenómenos de superficie libre en tanques, sobretodo en el mundo aeroespacial (Graham & Rodriguez, 1952; Verhagen & Van Wijngaarden, 1965; Abramson, 1966), el número de investigaciones sobre tanques estabilizadores pasivos y de diseños de ellos creció también (Stigter, 1996; Van Den Bosch & Vugts, 1966; Goodrich, 1968; Lewison, 1976). Recientemente, Moaleji & Greig (2007) analizó los distintos tipos de tanques y sus aplicaciones. ----------ABSTRACT---------- The hydrodynamics of liquid sloshing is not well understood. The wave loads, the wave impact event, the wave propagation and breaking are phenomena whose prediction and assessment are important for engineers. Nevertheless, these phenomena constitute unsolved problems for scientists and demand empirical research which is, when performed with scaled models, difficult to translate to prototypes. In this PhD thesis, experiments on two-dimensional sloshing have been performed, after calibrating the experimental facility of the CEHINAV. On the first hand, experiments on a model of an anti-roll showed the efficiency of such devices and the behavior of the liquid flows inside the container. Later, a more general investigation has been conducted on a section of a tank of an LNG tanker, excited with a rotatory motion around a fixed axis. 4 filling levels have been tested and pressure transducers have been placed at the still liquid levels. Highest waves have been observed for all the levels for excitation frequencies close to the first sloshing frequency predicted by means of the linear wave theory. These waves impact on the walls and highest pressures have been found for the lowest filling level. The impact of a breaking wave on a wall is a complex phenomena that has been studied in a third series of experiments. For the lowest filling level, varying the excitation period from 0.9 to 1.1 times the first sloshing period, great differences have been found in the way the breaking wave is formed and on the pressure resulting of its impact. The lowest air is entrapped during the breaking process, the highest the pressure peaks at the impact. Few numerical methods can deal with the phenomena aforementioned, due to the high deformation of the free-surface and the numerous fragmentation of the liquid. When using finite element codes, a special treatment of the free-surface is need, as for instance the Level Set (LS) technique, or the Volume of Fluid (VOF) method, increasing the numerical cost of the code. To solve these problems, meshless methods based on a Lagrangian formulation can be more effective than the grid-based methods. Among them, the socalled Smoothed Particle Hydrodynamics (SPH) method has been used in this thesis to simulate sloshing flows. The particular characteristics of SPH when aimed at simulating incompressible fluid have been reviewed. A SPH form of the compressible Navier-Stokes equations is obtained where the compressibility is controlled by imposing a low Mach number. At the end of this PhD thesis, the code used to simulate the impact of breaking wave was based on a normalized Gaussian function for the interpolation function. The free-slip boundary condition is imposed on rigid walls by use of the so-called Ghost Particles technique. A SPH viscous term has been used, whose correspondence with the Navier-Stokes viscosity has been demonstrated. The limitations of the code in terms of Reynolds number are discussed. They are due to the onset of numerical instabilities for high Reynolds numbers, if the number of fluid particles is not high enough. A filter has been applied periodically to the density field in order to smooth the unphysical numerical oscillations that appear due to the particles’ disorder. The Navier-Stokes equations have been integrated in time using a fourth-order Runge-Kutta scheme, allowing the use of higher time steps than the predictor corrector scheme initially used. Different ways to measure global moments and pressures on walls with SPH have been presented. The code has been validated comparing the numerical results with the experimental ones. The agreement is very good in terms of the evolution of the free-surface, including the breaking events and the impacts on walls. The moment created over the rolling axis compared also very well, showing the ability of SPH to deal with engineering problems such as the design of anti-roll tanks. Pressures resulting from the impacts of breaking waves on walls compare qualitatively well, but some improvements are needed, especially due to the bi-phasic character of this phenomena. No multiphasic SPH code has been developed in the time of this PhD thesis. SPH has finally been applied to simulated the internal flow in the coupled problem of the ship roll motion and the sloshing flow in an on-board tank. Since no valuable experimental data were available, the SPH results have been compared with non-linear potential theory, with empirical dissipative terms. Here again, the agreement between both methods is good, and experiments are under construction at the CEHINAV to provide useful experimental data for validation.