I. Introduction omputational fluid dynamics (CFD) codes available today provide access to high fidelity solutions for fluid flow problems, however turbulent combustion modeling is extremely challenging due to its inherent complex and transient nature. Furthermore, turbulence-chemistry interaction introduces more complication in modeling such problems. Large Eddy Simulation (LES) and detailed chemistry incorporated within CFD are popular choices among gas turbine combustor designers who are continuously seeking methods to improve accuracy of solution and reduce computational times. High efficiency markers available for propulsion systems are focusing more towards lean combustion, limited by understanding of flame stabilization1. LES solution explicitly solves for large scales of the flow, whilst small scales of motion are modeled. LES of reacting flow, interaction of turbulent mixing with chemical reactions initiates at the sub-grid (smallest) resolved scales and reaction rates at resolved scales are used for closure modeling. LES offers more accurate representation of complex unsteady processes, along with chemistry-turbulence interactions which results in more sophisticated understandings of fuel/air mixing, flame location, ignition, stabilization and Lean Blow Out (LBO). Reacting flow LES solution is significantly impacted from the selection of numerical schemes, Sub-Grid Scale turbulence, combustion chemistry and modeling of sub-grid scale turbulencechemistry interactions. The impact of CFD predictions from LES to reacting flows, have shown to be sensitive to physical setup and numerical parameters2. Thus, it has become increasingly important to assess CFD predictions of reacting flows in a systematic way such that best practice guidelines can be established. Complex flow structures interacting with flame such as shear layers, wakes, reaction zones and acoustic waves can be represented by performing computations on Bluff-Body-Stabilized-Flame. Volvo augmentor test rig3-4 has been studied numerically by other researchers through LES5-7. Such augmentor combustion simulations support predicting accuracy of reacting flow problems with special interest on phenomenon involving strong turbulence-chemistry interactions such as LBO, ignition and combustion instabilities. Experimental data for velocity, temperature and flow visualizations are used frequently by Model Validation for Propulsion (MVP) forum and its attendees. Two MVP workshops have so far been organized, with differences in thermal and acoustic boundary conditions. Current paper utilizes guidelines from MVP-2 workshop using ANSYS Fluent solver 8 for both structured and unstructured grids with high-order finite-volume CFD method; detailed chemistry is modeled with FGM using a finite rate based closure. Among many aforementioned challenges, this work focuses on detailed investigation of grid sensitivities including grid resolution, types of cell types and sensitivities due to boundary conditions. For all simulations, the other modeling attributes like chemistry resolution, turbulence-chemistry closure, discretization schemes etc., are not explored and same settings of all these parameters have been used for different grids to ensure precise comparison of solutions. In Section II, a description of the Volvo combustor experiment is detailed along with a summary of the computational setup. The results are discussed in two sections-section III and section IV. In section III, the focus is on structured grid where the computations are performed on cell size of 1 mm (fine grid), 2 mm (medium grid) and 4 mm (coarse grid), with and without Non-Reflecting Boundary Condition …