Analytic methods used to establish thermal radiation hazard safety boundaries from ignited hydrogen plumes are based on models previously developed for hydrocarbon jet fires. Radiative heat flux measurements of small- and medium-scale hydrogen jet flames (i.e., visible flame lengths < 10 m) compare favorably to theoretical calculations provided corrections are applied to correct for the product species thermal emittance and the optical flame thickness. Recently, Air Products and Chemicals Inc. commissioned flame radiation measurements from two larger-scale hydrogen jet flames to determine the applicability of current modeling approaches to these larger flames. The horizontally orientated releases were from 20.9 and 50.8 mm ID pipes with a nominal 60 barg source pressure and respective mass flow rates of 1.0 and 7.4 kg/s. Care was taken to ensure no particles were entrained into the flame, either from the internal piping or from the ground below. Radiometers were used to measure radiative heat fluxes at discrete points along the jet flame radial axis.
The estimated radiant fraction, defined as the radiative energy escaping relative to chemical energy released, exceeded correlation predictions for both flames. To determine why the deviation existed, an analysis of the data and experimental conditions was performed by Sandia National Laboratories’ Hydrogen Safety, Codes and Standards program. Since the releases were choked at the exit, a pseudo source nozzle model was needed to compute flame lengths and residence times, and the results were found to be sensitive to the formulation used. Furthermore, it was thought that ground surface reflection from the concrete pad and steel plates may have contributed to the increased recorded heat flux values. To quantify this impact, a weighted multi source flame radiation model was modified to include the influence of planar surface radiation. Model results were compared to lab-scale flames with a steel plate located close to and parallel with the release path. Relative to the flame without a plate, recorded heat flux values were found to increase by up to 50% for certain configurations, and the modified radiation model predicted these heat fluxes to within 10% provided a realistic steel reflectance value (0.8) was used. When the plate was heavily and uniformly oxidized, however, the reflectance was sharply attenuated. Model results that used the surface reflectance correction for the larger-scale flames produced good agreement with the heat flux data from the smaller of the two flames if an estimated reflectance of 0.5 was used, but was unable to fully explain the under predicted heat flux values for the larger flame.