Book Description

A comprehensive study of discontinuous finite element based high-order methods has been performed in this thesis, addressing a wide range of important issues related to high-order methods. The thesis starts with a detailed discussion of nodal based high-order methods and careful analysis of their stability properties. In particular, the formulations of nodal Discontinuous Galerkin method, Spectral Difference method, and Flux Reconstruction method for the scalar conservation laws are discussed first. The differences and similarities among these high-order schemes are carefully examined and effectively used to establish the linear stability of these methods. Stability proofs of nodal Discontinuous Galerkin method, Spectral Difference method, and Flux Reconstruction method subsequently lead to a new type of energy stable high-order scheme called Energy Stable Flux Reconstruction scheme. The extension of this new scheme from linear advection equation to the diffusion equation is formulated and discussed. The fundamental study of the high-order methods for scalar conservation laws lays the theoretical foundation for the subsequent extension to include conservation laws for fluid dynamics. The formulation of spectral difference method for the Navier-Stokes equations is first discussed. Validation tests to verify the resulting flow solver are presented. The extension of the spectral difference based Navier-Stokes flow solver from static fixed computational mesh to include dynamic moving deforming mesh is discussed next. An efficient mesh deformation algorithm that can handle substantial boundary movement is proposed and examined. The invariance of conservation laws mapping between coordinate systems allows the high-order scheme to be formulated on dynamic deforming meshes without deteriorating the formal order of accuracy of the underlying scheme. Detailed formulation, analysis, and validation results are presented. As a result of mesh deformation, the issue of geometric conservation needs to be addressed. The definition and origin of the geometric conservation law are discussed. The differential form of the geometric conservation law is derived from first principles for both the scalar conservation law and the fluid dynamic conservation laws. Subsequently a geometric conservative high-order scheme is formulated. The significance of geometric conservation on the stability and accuracy of the flow solution is examined. Finally a wide range of interesting fluid dynamic phenomena have been studied using the resulting high-order flow solver based on dynamic unstructured meshes. The representative test cases cover fluid dynamic phenomena ranging from completely laminar flows, to unsteady vortex dominated flows, and to flows exhibiting mixed regions of laminar, transitional, and turbulent structures. Other work that has been completed in this thesis is included in the appendix. In particular, continuous unsteady adjoint equations for advection and Burger's equations have been derived and solved using the high-order methods. The method of mesh deformation is reformulated as an optimization problem and used to achieve adaptive mesh refinement.