Hypersonic flows around complex geometries with adaptive mesh refinement and immersed boundary method

Abstract

This thesis develops and validates a computational fluid dynamics numerical method for hypersonic flows; and uses it to conduct two novel investigations. The numerical method involves a novel combination of structured adaptive mesh refinement, ghost-point immersed boundary and artificial dissipation shock-stable Euler flux discretisation. The method is high-order, low dissipation and stable up to Mach numbers $M \lesssim 30$ with stationary or moving complex geometries; it is shown to be suitable for direct numerical simulations of laminar and turbulent flows. The method's performance is assessed through various test cases. Firstly, heat transfer to proximal cylinders in hypersonic flow is investigated to improve understanding of destructive atmospheric entries of meteors, satellites and spacecraft components. Binary bodies and clusters with five bodies are considered. With binary proximal bodies, the heat load and peak heat transfer are augmented for either or both proximal bodies by $+20\%$ to $-90\%$ of an isolated body. Whereas with five bodies, the cluster-averaged heat load varied between $+20\%$ to $-60\%$ of an isolated body. Generally, clusters which are thin in the direction perpendicular to free-stream velocity and long in the direction parallel to the free-stream velocity have their heat load reduced. In contrast, clusters which are thick and thin in directions perpendicular and parallel to the free-stream velocity feel an increased heat load. Secondly, hypersonic ablation patterns are investigated. Ablation patterns form on spacecraft thermal protection systems and meteor surfaces, where their development and interactions with the boundary layer are poorly understood. Initially, a simple subliming sphere case without solid conduction in hypersonic laminar flow is used to validate the numerical method. Where the surface recession is artificially sped-up via the wall Damk\"{o}hler number without introducing significant errors in the shape change. Then, a case with transitional inflow over a backward facing step with a subliming boundary is devised. Differential ablation is observed to generate surface roughness and add vorticity to the boundary layer. A maximum surface recession of $\sim 0.8\times$ and a maximum surface fluctuation of $\sim 0.2\times$ the inflow boundary layer thickness were generated over two flow times.