Large eddy simulation of high speed convergent-divergent nozzle flows

Interest in developing a detailed understanding of jet plume aerodynamics has increased significantly in recent years, for both civil (noise reduction) and military (Infra-Red signature modelling) aerospace applications. Such flows are critically dependent on turbulence modelling of the jet plume shear layer mixing. Reynolds averaged Navier Stokes (RANS) CFD tends to overpredict while Large Eddy Simulation (LES) CFD underpredicts potential core length. Difficulties in LES begin with the challenge of providing accurate resolution of thin turbulent boundary layers at nozzle exit. Providing physically meaningful 3D unsteady LES inlet conditions is a challenge in nozzle flows since turbulence at nozzle inlet experiences relaminarisation, which determines the boundary layer state at nozzle exit. The present thesis addresses these challenges by developing and validating against benchmark measurements an LES approach for nozzle/plume flows based on an advanced inlet condition treatment and an improved level of Sub-Grid-Scale (SGS) modelling. A technique for synthetic inlet condition generation based on a rescaling/recycling method (R2M) for LES predictions of nozzle flows has been applied and validated in the present work. Results reveal the benefits of this method such that self-consistent, correlated turbulent structures were sustained throughout the high acceleration region associated with nozzle convergence, with the turbulence anisotropy developing in the expected manner. The LES results for velocity profile shape at nozzle exit are better than low Re RANS predictions. Use of the Smagorinsky SGS closure produced level of turbulence energy at nozzle exit significantly lesser than measured. A recently proposed SGS model by Piomelli and Guerts (PGSGS) that defines the SGS length scale based on local turbulence quantities using a mesh independent formulation was also applied to the nozzle flow test case with significant improvement in the turbulence energy development through the nozzle. The LES method is applied to a supersonic jet discharging from a rectangular convergentdivergent nozzle. Results show that the R2M technique was able to generate realistic turbulence conditions at nozzle inlet that were consistent with available measured data. Using a carefully designed mesh and the advanced PGSGS model, turbulent structures were sustained through the nozzle, enabling good prediction of the nozzle exit boundary layer state and near field development. The improved capture of shear layer turbulence enabled better predictions of shear layer growth, leading to improved capture of shock cell behaviour and potential core length.