heal.abstract |
The understanding of the motion of water waves is of fundamental importance for many
applications related to disciplines such Naval and Marine Hydrodynamics, Coastal
and Environmental Engineering, and Oceanography. Even under the simplifying
assumptions that fluid is ideal and the flow irrotational, the complete mathematical
formulation of the free-boundary problem of water waves is very complicated and its
theoretical and numerical study comprises a contemporary direction of research.
In the first part of this thesis, a new system of two Hamiltonian equations is derived,
governing the evolution of free-surface waves. This system is coupled with a timeindependent
Coupled Mode System (CMS), called the substrate problem, that accounts
for the internal fluid kinematics. The derivation is based on the use of Luke’s
variational principle in conjunction with an appropriate series representation of the
velocity potential. The critical feature of this approach, initiated in (Athanassoulis &
Belibassakis 1999, 2000), is the use of an enhanced vertical modal expansion that
serves as an exact representation of the velocity potential in terms of horizontal modal
amplitudes. Herein, we study further and justify this expansion. In particular, it is
proved that under appropriate smoothness assumptions, the modal amplitudes exhibit
rapid decay, ensuring that the infinite series can be termwise differentiated in the nonuniform
fluid domain, including its physical boundaries. This justifies the variational
procedure and proves that the resulting system, called Hamiltonian/Coupled-Mode
System (HCMS), is an exact reformulation of the complete hydrodynamic problem,
and therefore it is valid for fully nonlinear waves and significantly varying seabeds.
In fact, it is a modal version of Zakharov/Craig-Sulem Hamiltonian formulation
(Zakharov 1968, Craig & Sulem 1993) with a new, versatile and efficient representation
of the Dirichlet to Neumann operator (DtN) operator, needed for the closure of the
non-local evolution equations. No smallness assumptions are made, that is, the
present approach is a non-pertubative one. In HCMS, the DtN operator is defined
in terms of one of the unknown modal amplitudes, namely, the free-surface modal
amplitude. Its computation avoids the numerical solution of the Laplace equation in
the whole fluid domain, required in direct numerical methods, and the evaluation
of higher-order horizontal derivatives, required in Boussinesq or other higher-order
pertubative methods. Instead, a system of horizontal second order partial differential
equations needs to be solved.
In the second part of the thesis, our theoretical results are exploited for the numerical
solution of various nonlinear water wave problems. The backbone of our numerical method is the computation of the DtN operator through its modal characterization
which is achieved by a fourth-order finite-difference method. The accuracy and
convergence of the new characterization of the DtN operator is assessed in test cases
of highly non-uniform domains and our theoretical findings concerning the rate of
decay of the modal amplitudes are numerically verified. This preparatory investigation
demonstrates that a small number of modes suffices for the accurate computation of
the DtN operator even in extremely deformed domains. Subsequently, a number of
physically interesting water-wave problems, over flat as well as varying bathymetry,
are considered. The first application concerns the computation of steady travelling
periodic waves above a flat bottom for a wide range of nonlinearity and shallowness
conditions up to the breaking limit. Next, we turn to the time integration of the
new evolution equations by employing a fourth-order Runge-Kutta for the simulation
of wave interactions with variable bathymetry and vertical walls. Computations
are validated against predictions from laboratory experiments and other numerical
methods in connection with several nonlinear phenomena. In particular we study
the interaction of solitary waves with a vertical wall (reflection) and a plane beach
(shoaling) and the transformation of regular incident waves past submerged obstacles
(harmonic generation) or undulating bathymetry (Bragg scattering). Numerical
results on the interaction of a solitary wave with an undulating bottom patch are also
provided. The present method provides stable and accurate long time simulations
of nonlinear waves in various depths, from deep to shallow waters, avoiding the
computational burden of direct numerical methods as well as the use of filtering
techniques, frequently required in pertubative approaches. |
en |