University of Notre Dame
Browse
SeboldtA052022D.pdf (6.15 MB)

Numerical Methods and Analysis for Fluid-Structure Interaction Problems

Download (6.15 MB)
thesis
posted on 2022-05-19, 00:00 authored by Anyastassia Seboldt

Fluid-structure interaction (FSI) problems help to describe many physical phenomena in our world, including topics such as aerodynamics, hydrodynamics, geophysical flows, engineering, and biomedical applications. The problems may be mathematically modeled using a system of Navier-Stokes and elastodynamics partial differential equations (PDEs). The fluid of interest is viscous, incompressible, and Newtonian, and the solid is elastic and thick, meaning the solid may be described in the same dimension as the fluid. When the densities of the fluid and solid are comparable, an added-mass effect is introduced where the force from the fluid acts as a mass on the fluid-structure interface. This added complexity makes finding solutions to this two-way coupled problem especially difficult to numerically solve.

In order to solve these FSI problems, three novel algorithms are presented and analyzed. More specifically, the solvers developed throughout this dissertation are based on partitioning, using both loosely and strongly coupled techniques. The proposed schemes are then analyzed for stability and convergence and their performance is investigated in numerical examples. One of the main novelties of each of the three schemes is based on the implementation of generalized Robin-like boundary conditions, which are imposed on each subproblem in order to account for matching velocities and stresses at the fluid-structure interface. The adaptation of our method compared to classical Robin-Robin boundary conditions allows us to use energy estimates to prove our algorithm applied to a moving domain problem is unconditionally stable.

The time-discretization varies between methods and spatial discretization is done using the finite element method (FEM). When considering spatial discretization, it is important to note that a saddle-point problem arises due to the fluid’s incompressibility. Hence, certain conditions of the finite element spaces are met, which is addressed in our choices of finite elements used in the numerical examples.

History

Date Modified

2022-08-02

Defense Date

2022-04-19

CIP Code

  • 27.9999

Research Director(s)

Martina Bukač

Committee Members

Daniele Schiavazzi Yongtao Zhang

Degree

  • Doctor of Philosophy

Degree Level

  • Doctoral Dissertation

Alternate Identifier

1338037093

Library Record

6263175

OCLC Number

1338037093

Program Name

  • Applied and Computational Mathematics and Statistics

Usage metrics

    Dissertations

    Categories

    No categories selected

    Keywords

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC