An Adaptive Finite Element Method For Particle Suspension In Viscoelastic Fluid

Author : Ernesto Castillo del Barrio
Publisher :
ISBN 13 :
Total Pages : 214 pages
Book Rating : 4.:/5 (112 download)

DOWNLOAD NOW!

Download or read book Stabilized Finite Element Formulations for the Three-field Viscoelastic Fluid Flow Problem written by Ernesto Castillo del Barrio and published by . This book was released on 2016 with total page 214 pages. Available in PDF, EPUB and Kindle. Book excerpt: The Finite Element Method (FEM) is a powerful numerical tool, that permits the resolution of problems defined by partial differential equations, very often employed to deal with the numerical simulation of multiphysics problems. In this work, we use it to approximate numerically the viscoelastic fluid flow problem, which involves the resolution of the standard Navier-Stokes equations for velocity and pressure, and another tensorial reactive-convective constitutive equation for the elastic part of the stress, that describes the viscoelastic nature of the fluid. The three-field (velocity-pressure-stress) mixed formulation of the incompressible Navier-Stokes problem, either in the elastic and in the non-elastic case, can lead to two different types of numerical instabilities. The first is associated with the incompressibility and loss of stability of the stress field, and the second with the dominant convection. The first type of instabilities can be overcome by choosing an interpolation for the unknowns that satisfies the two inf-sup conditions that restrict the mixed problem, whereas the dominant convection requires a stabilized formulation in any case. In this work, different stabilized schemes of the Sub-Grid-Scale (SGS) type are proposed to solve the three-field problem, first for quasi Newtonian fluids and then for solving the viscoelastic case. The proposed methods allow one to use equal interpolation for the problem unknowns and to stabilize dominant convective terms both in the momentum and in the constitutive equation. Starting from a residual based formulation used in the quasi-Newtonian case, a non-residual based formulation is proposed in the viscoelastic case which is shown to have superior behavior when there are numerical or geometrical singularities. The stabilized finite element formulations presented in the work yield a global stable solution, however, if the solution presents very high gradients, local oscillations may still remain. In order to alleviate these local instabilities, a general discontinuity-capturing technique for the elastic stress is also proposed. The monolithic resolution of the three-field viscoelastic problem could be extremely expensive computationally, particularly, in the threedimensional case with ten degrees of freedom per node. A fractional step approach motivated in the classical pressure segregation algorithms used in the two-field Navier-Stokes problem is presented in the work.The algorithms designed allow one the resolution of the system of equations that define the problem in a fully decoupled manner, reducing in this way the CPU time and memory requirements with respect to the monolithic case. The numerical simulation of moving interfaces involved in two-fluid flow problems is an important topic in many industrial processes and physical situations. If we solve the problem using a fixed mesh approach, when the interface between both fluids cuts an element, the discontinuity in the material properties leads to discontinuities in the gradients of the unknowns which cannot be captured using a standard finite element interpolation. The method presented in this work features a local enrichment for the pressure unknowns which allows one to capture pressure gradient discontinuities in fluids presenting different density values. The stability and convergence of the non-residual formulation used for viscoelastic fluids is analyzed in the last part of the work, for a linearized stationary case of the Oseen type and for the semi-discrete time dependent non-linear case. In both cases, it is shown that the formulation is stable and optimally convergent under suitable regularity assumptions.