Resumen de Adaptive mesh simulations of compressible flows using stabilized formulations
This thesis investigates numerical methods that approximate the solution of compressible flow equations.
The first part of the thesis is committed to studying the Variational Multi-Scale (VMS) finite element approximation of several compressible flow equations. In particular, the one-dimensional Burgers equation in the Fourier space, and the compressible Navier-Stokes equations written in both conservative and primitive variables are considered. The approximations made for the VMS formulation are extensively researched; the design of the matrix of stabilization parameters, the definition of the space where the subscales live, the inclusion of the temporal derivatives of the subscales, and the non-linear tracking of the subscales are formulated. Also, the addition of local artificial diffusion in the form of shock capturing techniques is included. The accuracy of the formulations is studied for several regimes of the compressible flow, from aeroacoustic flows at low Mach numbers to supersonic shocks.
The second part of the thesis is devoted to make the solution of the smallest fluctuating scales of the compressible flow affordable. To this end, a novel algorithm for $h-$refinement of computational physics meshes in a distributed parallel setting, together with the solution of some refinement test cases in supercomputers are presented. The definition of an explicit a-posteriori error estimator that can be used in the adaptive mesh refinement simulations of compressible flows is also developed; the proposed methodology employs the variational subscales as a local error estimate that drives the mesh refinement.
The numerical methods proposed in this thesis are capable to describe the high-frequency fluctuations of compressible flows, especially, the ones corresponding to complex aeroacoustic applications. Precisely, the direct simulation of the fricative [s] sound inside a realistic geometry of the human vocal tract is achieved at the end of the thesis.
