Large-eddy simulation of turbulent swirling flame in a dual swirl H2-air coaxial injector

Thumbnail Image
Journal Title
Journal ISSN
Volume Title
Insinööritieteiden korkeakoulu | Master's thesis
Mechanical Engineering
Degree programme
Master's Programme in Mechanical Engineering (MEC)
This study presents a numerical framework to validate the stabilized swirling attached flame (Flame-A) obtained from the HYdrogen LOw NOx laboratory scale coaxial dual-swirl injector (HYLON) using the open-source CFD code OpenFOAM integrated with the dynamic load balancer chemistry library DLBFoam. The investigation explores modelling choices in both non-reactive and reactive scenarios. For the non-reactive case, three studies are presented: firstly the effect of the mesh refinement configuration; secondly the effect of the turbulence model including unsteady Reynolds averaged (URANS) and large-eddy simulation (LES) with two distinct explicit subgrid models (SGS); and finally the effect of the discretization schemes flux limiter. All of the studies are presented based on the quantitative comparison with the experimental particle image velocimetry (PIV) data. In the reactive scenario, the evaluation of various recent chemical kinetic mechanisms in finite-rate chemistry simulations for pure hydrogen is initially discussed, using one-dimensional (1D) free propagating flame modelling. Consequently, a quantitative and qualitative validation is carried out of the three-dimensional (3D) H2 flame in terms of velocity profiles, pressure drop, outlet temperature, flame heat release and NOx emissions. The effect of the flame on the flow main structures is examined by comparing the non-reacting and reacting cases and observing the change in the central and outer recirculation zones (CRZ and IRZ) while comparing the findings from the numerical results to the reference experimental data. The analysis of the pure hydrogen attached flame characteristics reveals that the flame is forming in a non-premixed mode which explains the ability of the perfectly stirred reactor (PSR) model to predict the flame relatively accurately and align with previous observations in literature.
Vuorinen, Ville
Thesis advisor
Tamadonfar, Parsa
computational combustion, swirling flow, large-eddy simulation, hydrogen
Other note