Multi-Resolution Analysis of Subgrid Turbulence / Chemistry Interactions in a Supersonic Hydrogen-Air Diffusion Flame

Abstract

The influences of turbulence-chemistry interactions (TCIs) and subgrid-scale (SGS) transport effects on the prediction of the structure of a hydrogen / vitiated-air diffusion flame are studied in this work using a multi-resolution analysis (MRA) technique. MRA couples sequences of large-eddy simulations at different mesh levels using a variety of constraining source terms. Eddy structures on the constrained coarser meshes closely track those of the finest mesh, enabling direct examination of the role of coarse-graining on the flow structure and on the ability of subgrid models to capture the effects of unresolved information. A fifteen-mesh MRA hierarchy developed over three mesh levels is used as the technique for evaluating the performance of laminar chemistry (LC), least-squares minimization (LSM), and partially-stirred reactor (PaSR) TCI models. A locally-evolving sample space constructed from subgrid Reynolds (Re) and Damköhler (Da) numbers is used to help quantify the actions of the selected TCI models – the structure of the joint PDF and conditional statistics of various reactivity measures supply this information. The results show similar levels of predictive capability for LC and LSM models constrained by the fine-mesh data but indicate a modest level of performance improvement for LSM applied to independent, coarser-mesh realizations of the flame. The tested PaSR model fails to predict substantial heat release. Information contained in the joint Re-Da PDFs constructed using different types of data is also used to determine an optimal model constant for a scale-similarity parameterization of a rms subgrid velocity.