dc.description.abstract |
This thesis project focuses on the numerical solutions of selected nonlinear hyperbolic sys
tems of partial differential equations (PDEs) describing incompressible and compressible
flows. Such type of PDEs are used to simulate various flows in science and engineering.
The underlying physics of such systems of PDEs is very complex and some mathematical
and computational issues are associated with them. For instance, they may contain non
conservative terms or may be weakly hyperbolic. The strong nonlinearity of the systems
could generate sharp fronts in the solutions in a finite time interval, even for smooth initial
data. Moreover, accurate discretization of the non-conservative terms is a challenge task
for the numerical solution techniques. In the presence of non-conservative terms, well
balancing, positivity preservation and capturing of steady states demand special attention
during the application of a numerical algorithm. In this thesis project, we develop exact
Riemann solvers for the one-dimensional Ripa model, containing shallow water equations
that incorporate horizontal temperature gradients and considering both flat and non flat
bottom topographies. Such Riemann solvers are helpful for understanding the behavior of
solutions, as these solutions contain fundamental physical and mathematical characters of
the set of conservation laws. Such solvers are also very helpful for evaluating performance
of the numerical schemes for more complex models. Afterwards, third order well-balanced
finite volume weighted essentially non-oscillatory (FV WENO) schemes are applied to solve
the same model equations in one and two space dimensions and a Runge-Kutta discontin
uous Galerkin (RKDG) finite element method is applied to solve this model in one space
dimension. In the case of compressible fluid flow models, an upwind conservation element
and solution element (CE/SE) method and third order finite volume WENO schemes are applied to solve the dusty gas and two-phase flow models. The suggested numerical schemes
are able to tackle the above mentioned associated difficulties in a more efficient manner.
The accuracy and order of convergence of the proposed numerical schemes are analyzed
qualitatively and quantitatively. A number of numerical test problems are considered and
results of the suggested numerical schemes are compared with the derived exact Riemann
solutions, results available in the literature, and with the results of a high resolution central
upwind (CUP) scheme. |
en_US |