Metric based mesh adaptation methodology applied to HEMLAB algorithm

Abstract

This dissertation presents a detailed study on metric-based mesh adaptation methodologies applied to high-lift aerodynamic simulations within the HEMLAB framework. The research integrates advanced computational techniques to enhance the accuracy and efficiency of computational fluid dynamics (CFD) solvers, particularly for high-lift configurations analyzed in the AIAA High Lift Prediction Workshop Series (HLPW). The primary focus is on improving numerical predictions by refining computational meshes dynamically in response to flow characteristics. A key contribution of this research is the development and integration of metric-based anisotropic mesh adaptation strategies. These techniques refine computational grids in regions of high flow gradients, such as boundary layers and wake structures, ensuring improved resolution without excessive computational cost. The numerical studies conducted on HLPW-3, HLPW-4, and HLPW-5 test cases illustrate the impact of these adaptation methods on aerodynamic loads, particularly in improving lift and pressure coefficient distributions. In addition to mesh adaptation, this study incorporates solution methodologies through the integration of the Scalable Nonlinear Equations Solvers (SNES) framework. By utilizing nonlinear Newton-Krylov methods with efficient preconditioners the computational cost of solving nonlinear equations is significantly reduced. The accurate Jacobian evaluation as a preconditiner further enchance solver efficiency. Turbulence modeling is another aspect of this research. The initial version of HEMLAB was limited to the standard Spalart-Allmaras (SA) model. This study extends its turbulence modeling capabilities by incorporating SA-neg, SA-neg-ft2, and SA-QCR-2000 models. These modifications improve the solver's accuracy and convergence in predicting separated flows, transitional effects, and streamline curvature influences. A second-order discretization scheme is also introduced to the SA model, enhancing numerical precision and reducing diffusion errors. Furthermore, this thesis highlights the significance of machine precision in high-fidelity simulations. The methodologies developed in this work contribute to the broader field of CFD by providing improved numerical accuracy, efficiency, and adaptability. Future research directions include refining transition modeling techniques, extending adaptive meshing strategies such as goal based adaptation strategies and improving sensor function and further investigation of cross diffusion effects on highly anisotropic meshes for the aircraft configurations.