Modelización del cambio de fase sólido-líquido. Aplicación en sistemas de acumulación de energía térmica

  1. Vidal Jiménez, Bárbara
Dirixida por:
  1. Miquel Costa Pérez Director
  2. Asensio Oliva Llena Director
  3. Carlos David Pérez Segarra Director

Universidade de defensa: Universitat Politècnica de Catalunya (UPC)

Fecha de defensa: 10 de setembro de 2007

Tribunal:
  1. José Manuel Pinazo Ojer Presidente/a
  2. Jesús Castro Gonzalez Secretario/a
  3. Pere Colomer Vilanova Vogal
  4. Antonio Pascau Benito Vogal
  5. José Fernández Seara Vogal

Tipo: Tese

Teseo: 138003 DIALNET lock_openTDX editor

Resumo

In this thesis a detailed numerical simulation of liquid-solid phase change phenomena has been made, because this phenomenology is of great interest in different industrial areas. The simulation done implies problems of nonlinearity, strong couplings and movable interphase. Like a result, only for the simplest configurations analytical tools can be used, whereas to solve the most of interest problems numerical methods are needed. These methods consist in the discretisation of the equations that define the phenomenology in small cells or control volumes. In this study the Finite Volume Method (FVM) has been used for the governing equations discretisation using Cartesian meshes. A displaced mesh is used; this means that the components of the speed vector are calculated in the faces of the control volumes, which allows a correct coupling between the continuity and momentum equations. So the code as the numerical solutions have been properly verified. The code verification consist of verifying that this is free of programming errors and that the behaviour of the numerical schemes is agreed with the theoretical one. For the verification of the numerical solution the Richardson Extrapolation Method or a mesh refinement study have been used. Once the code and the numerical solutions have been properly verified, the final validation of the process simulation is obtained comparing the numerical results with experimental ones. The Gallium melting problem in a square cavity heated by a side has been widely used by investigators with the objective of evaluatingthe numerical methods used for solving the phase change phenomena. Although this material have the advantages of its thermophysical properties are well established, the phase change temperature is near the room temperature and is a material with a great industrial interest, also presents some disadvantages like having an anisotropic thermal conductivity of the solid phase. However, the great number of experimental works that can been found in the literature, lead us to choose this material for doing a detailed study of the solid liquid phase change in this work. Although in the literature this problem appears with different configurations, in this thesis we have centered in the study of the case with an aspect ratio (height/width) of 0.5. Different numerical methods exist to solve solid-liquid phase change problems: methods that follow the moving interphase, methods that fix the moving interphase, etc. In this work we have used the Enthalpy Method because it allows us to use a fixed mesh in all the domain, the Stefan condition is imposed implicitly, it allows the coexistence of more than one front of phase change and allows the interphase has a thickness. This case has the singularity of being in a low Prandtl number range. This causes that a turbulent state has been reached for relatively low Rayleigh numbers. This has made think us on the convenience of making a detailed study to determine the transition of permanent regime to oscillating flow and from oscillating flow to chaotic regime. The problems than have been found in the accomplishment of this work have lead us to the use of multiblock method. This method is used considering incompressible flows and moved meshes. We will explain the modifications that have been necessary for using this method in the phase change phenomena. Two approaches have been used: a conservative one and a pressure based one. For obtaining the reference solution has been use the pressure based method, because the conservative method presents discrepancies between the obtained solution with an only subdomain and the obtained with several subdomains. Finally, a parametric study has been done, considering different aspect ratios, boundary conditions and variations of ±10% in the thermophysical properties with respect to the reference case, with the objective to see as these changes affect on the phenomenology that occupies to us.