Modeling and simulation of V-Gutter stabilized turbulent flames

Travis Calob Fisher, Purdue University

Abstract

Bluff bodies are used to stabilize flames in high velocity cross flow, and are commonly utilized in jet engine afterburners, ramjets, and scramjets. The purpose of this work is to examine the non-reacting and reacting flow dynamics in a V-Gutter bluff body configuration that can be predicted using the commercially available Computational Fluid Dynamics (CFD) package Fluent, and gage the ability of the models employed to predict flame blow out, or global extinction of the flame. Non-reacting flow was simulated using Large Eddy Simulations (LES) and Detached Eddy Simulation (DES), or modeled using Unsteady Reynolds Averaged Navier Stokes (URANS). Reacting flow was simulated using LES or URANS with the Finite Rate, Eddy Dissipation (FRED) model for turbulence-chemistry interactions. To address the ability of the solver to predict flame blow out, a parametric study was conducted in two dimensions for different turbulence-chemistry interaction models and chemical kinetic mechanisms. Non-reacting LES resulted in a good match to experimental data. LES matched the measured recirculation zone length in the wake of the bluff body and the downstream mean velocity better than the other turbulence models. URANS simulations did not accurately reproduce the mean flow field. It was found that the Realizeable k-ε model failed to predict large scale unsteadiness, which is known to occur in this configuration, but the RNG k-ε model was able to predict this unsteadiness. Reacting LES and URANS simulations exhibited a poor match to experimental data. The reacting LES result was examined in more detail, and characteristics of reacting flow from theory are confirmed. The local Damk-Nohler number field was calculated to characterize local flame stability. The parametric study found that for global and two-step propane-air chemistry, URANS does not accurately predict lean blow out for any of the combustion models tested. Flame extinction was predicted using the Eddy Dissipation Concept (EDC) model with a more detailed chemical mechanism at the blow out equivalence ratio. This work has suggested that for bluff body stabilized flames, it is not practical to perform phenomenological studies of flame dynamics using this commercial package. Poor parallel scaling limited the use of LES to simple combustion models and few model revisions. The work does indicate, however, that URANS calculations can be conducted using Fluent with more detailed chemistry and appropriate turbulence-chemistry interaction models to impact the design of bluff body configurations.

Degree

M.S.M.E.

Advisors

Frankel, Purdue University.

Subject Area

Mechanical engineering

COinS