Numerical Investigation of Lean Hydrogen-Air Flame Stabilization Regimes Using Large-Eddy Simulations

Abstract

Abstract High-fidelity large-eddy simulations (LES) of turbulent combustion in a typical H2-air flame have been conducted in this work. The simulations are based on a laboratory-scale coaxial dual-swirl injector in which fuel and air are injected separately. Based on the thermal power, at a given equivalence ratio, two flame archetypes are observed in the experiments: a flame anchored to the injector (at ∼4 kW) and an aerodynamically stabilized flame (at ∼10 kW). LES allows the scrutiny of the flame structure and stabilization mechanisms seen with each archetype. The LES is first validated on a non-reacting configuration using Particle Image Velocimetry (PIV), followed by extending the modeling formulation to account for combustion. The modeling formulations successfully retrieve the two regimes, and the stabilization mechanism is investigated further by comparing the LES results with PIV and OH* chemiluminescence images. The mean velocity field for both operating conditions suggests the existence of a strong central recirculation zone (CRZ), which penetrates along the central axis in the injector nozzles, leading to a radial deviation of the central hydrogen jet. As a result, CRZ favors the stabilization of one mode over the other. In one mode, the flame anchors over the lip of the injector and exists along the mixing layer between hydrogen and swirling air jets. In the latter mode, the flame is aerodynamically lifted and stabilizes in the inner shear layer between CRZ and exiting swirling air jets. Further, LES analysis suggests that the flame is diffusion-dominated in the anchored regime, while the lifted regime has partial premixing and diffusion fronts. Another aspect of numerical investigations is to predict the NOx generated for each operating point. The combustion model formulation is further extended to account for NOx by using the extended Zeldovich mechanism. Finally, various approaches are compared for computational efficiency to predict different NOx levels for each operating condition. In general, LES results successfully predict the flame stabilization regimes along with NOx.