Ph.D. Thesis Defense by

Balaji Muralidharan

(Advisor: Prof. Suresh Menon)

“Towards Simulations of Turbulent Reacting Flow Problems in Complex Geometries with Moving/Stationary Boundaries”

Friday, 28 April @11:00 a.m.
Montgomery Knight Building Room 317

Abstract:
Many useful engineering devices involve moving boundaries interacting with a reacting compressible flow. Examples of such applications include propulsion systems with moving components such as Internal Combustion (IC) engines, hypersonic propulsive devices such as Oblique Detonation Wave (ODW) engines and solid rocket motors involving regressing propellant surfaces. Computational Fluid Dynamics (CFD) can be effectively employed to study these systems. However, conventional numerical methods face several difficulties related to grid generation, treatment of moving boundaries, lack of adequate grid resolution at an affordable computational cost, and shortcomings in closure models required for Large Eddy Simulation (LES). This thesis demonstrates new accurate numerical models and subgrid closures for LES of problems in non-trivial geometries with moving boundaries.  A new high-order adaptive cut-cell based embedded boundary method is developed for viscous flows, which can provide a smooth and accurate reconstruction to predict the near-wall shear stress and pressure distribution. The method can achieve a high order of accuracy even under adverse geometrical constraints such as narrow gaps and sharp corners due to a novel and robust cell clustering algorithm. This algorithm also enforces the stability of the numerical scheme in the presence of arbitrary low volume cells formed in the cell cutting process.  Additionally, an extended cell clustering approach, which can achieve exact conservation of mass, momentum, and energy is proposed for moving boundaries.

The embedded boundary method is built on a massively parallel framework that performs block structured Adaptive Mesh Refinement (AMR) by interfacing with the BoxLib open source library.  This modeling framework is then applied to study fundamental physics in high-speed propulsion systems, for example, shock-turbulence interactions, flame-turbulence interaction, and flame/detonation stabilization in a reacting system.  LES using the multilevel subgrid closure for flow and chemistry is used to study flame anchoring in a transverse reacting jet in cross flow. Important mechanisms that stabilize the flame are identified and shown to be consistent with past observations from experiments and using direct numerical simulations (DNS) but obtained here using much coarser grid LES. Finally, to demonstrate the ability of the methodology to simulate moving bodies in a reactive system, DNS of a hypersonic projectile fired into a reacting flow is performed to reveal key effects of pressure on the stabilization of detonation ahead of the projectile.  

Committee:
Prof. Stephen M. Ruffin (AE)
Prof. Lakshmi N. Sankar (AE)
Prof. Julian J. Rimoli (AE)
Prof. Alexander Alexeev (ME)