Modeling of large-scale under-expanded hydrogen jet fires

Abstract

Large-scale hydrogen under-expanded jet flames and lab-scale H2 and CH4 jet flames are simulated by using the standard k-ε model coupled to a hybrid flamelet/transported PDF method and a narrow band correlated-k gas radiation model. The set of flames considered covers a wide range of residence times and optical thicknesses. The notional nozzle concept is adopted to determine injection conditions for the large-scale chocked flames. Model predictions in terms of flame geometry, flame structure, radiant fraction and radiative flux are consistent with the experimental data whatever the scales. Model results show that these flames do not verify the optically-thin approximation since the part of emitted radiant power re-absorbed within the flame ranges from 11% for the smallest H2 flame to about 70% for the largest H2 flame. Neglecting the turbulence-radiation interaction underestimates significantly the radiant fraction whatever the scale and these discrepancies are enhanced with increasing residence time and optical thickness. A simple analysis based on the assumption of homogenous flame is used to correlate the experimental radiant fraction as a function of τGEm(1−Q˙abs/Q˙em) where τG, Em and Q˙abs/Q˙em represent the residence time, an equivalent emission term and the part of emitted radiant power re-absorbed within the flame. Model results are used to provide a proper estimation of the equivalent emission term and self-absorption.