Abstract:
The main subject of this thesis is to analyze the incompressible Navier-Stokes-Nernst-Planck-Poisson system for bounded domains. Such a system is used as a model in electrohydrodynamics or physicochemical models. First, we verify existence of weak and strong solutions. Moreover, we are able to characterize the weak solutions by an energy and an entropy law. The concentrations in the Nernst-Planck equations additionally are non-negative and bounded.
These results motivate to construct convergent space-time discretizations based on low order finite elements, where solutions of the discrete problem preserve the characteristic properties from the continuous context.
For this purpose, we first introduce an energy based and an entropy based
approximation for the simpler Nernst-Planck-Poisson sub-system which is also called the van Roosbroeck equations in the semiconductor theory. The main focus is to study qualitative properties of the two discretization strategies at finite discretization scales, like conservation of mass, non-negativity, discrete maximum principle, decay of discrete energies and entropies to study
long-time asymptotics. The energy based scheme uses the M-matrix property to prove non-negativity and boundedness of iterates. Here, we have to assume more regular initial data in order to verify a perturbed entropy law.
This deficiency for the entropy behavior is resolved by an entropy based scheme allowing for an entropy inequality without any additional assumptions. However, in turn, the latter scheme suffers from weaker results, such as quasi non-negativity, and the lack of a discrete maximum principle.
These results suggest to follow the energy based approach for the coupled incompressible Navier-Stokes-Nernst-Planck-Poisson system. The main obstacle here is the lack of regularity of velocity fields from the Navier-Stokes
equations which makes the verification of the M-matrix property in the Nernst-Planck-Poisson part more difficult. We therefore regularize the discrete momentum equation by an additional term such that the incompressible Navier-Stokes equations arise as the limit of the discrete problem.
Main results then include non-negativity, conservation of mass, and a discrete maximum principle for concentrations, and a discrete energy and (in two dimensions) a discrete entropy law for iterates which solve a nonlinear algebraic problem: A fixed-point scheme is introduced for both, theoretical and practical purposes to solve the nonlinear problem together with an appropriate stopping criterion. Overall convergence of solutions to weak solutions of the incompressible Navier-Stokes-Nernst-Planck-Poisson equations is shown. We conclude with the verification of optimal convergence rates for a suggested time-splitting scheme whose iterates converge to (locally existing) strong solutions of the electrohydrodynamic system.
At the end we compare the energy based scheme and the splitting scheme by
numerical experiments.