Master thesis and internship[BR]- Master's thesis : Implementation of a semi-inverse coupling method in the viscous-inviscid interaction code BLASTER[BR]- Integration Internship

Abstract

The aeronautical industry faces significant challenges in reducing fuel consumption to lower CO2 emissions and operating costs. Computational fluid dynamics (CFD) plays a vital role in modern aircraft design by numerically solving flow equations. However, the high computa- tional cost of methods like Direct Numerical Simulation (DNS) and Large Eddy Simulation (LES) makes them impractical in early design stages. While Reynolds-averaged Navier- Stokes (RANS) simulations are widely used, they remain resource-intensive, necessitating alternative approaches. The viscous-inviscid interaction (VII) technique offers a promising solution by combining an inviscid solver with a viscous boundary layer solver, accounting for fluid viscosity at a lower computational cost than RANS. Despite its advantages, traditional VII methods encounter challenges, particularly the occurrence of the Goldstein singularity in adverse pressure gradient flows, causing convergence issues. This thesis focuses on enhancing a VII code, BLASTER, by implementing a semi-inverse coupling method. BLASTER integrates an unstructured full-potential finite element solver for inviscid flow with an integral boundary layer solver using a quasi-simultaneous coupling approach, which performs well in many scenarios but struggles with largely separated flows or laminar separation, which might occur at low Reynolds numbers and high angles of attack. These limitations are assumed to stem from Goldstein’s singularity, which the quasi- simultaneous method is unable to fully overcome. The semi-inverse coupling method avoids this singularity but converges slowly. To address this, the thesis combines both methods, dynamically selecting the coupling approach based on local boundary layer conditions. After deriving the equations for laminar and turbulent boundary layers, the semi-inverse coupling and hybrid algorithm were implemented. Simulations on various flow cases showed that while the semi-inverse method performs well for simple flows, it does not significantly extend BLASTER’s applicability. The combined algorithm further limits the range of treat- able flows.