The N-GAS model for predicting smoke toxicity has been proposed for many years by NIST.However,almost all the existing CFD software cannot accurately predict the toxicity of smoke using the model because of the absence of the toxic gas concentration of HCN,NO x,HCl and HBr.In this work,an approach for predicting fire smoke toxicity was developed and demonstrated.A detailed mechanism including these fire effluents was constructed firstly,and the subsequent generation of state relationship among fire effluents,mixture fraction and strain rate was conducted by using opposed-flow flame technique.A mixture fraction-based combustion model used in FDS code was modified,and meanwhile the scalar dissipation rate transport equation was numerically solved.Thus the concentration of fire effluents as the function of mixture fraction and scalar dissipation rate can be calculated through a look-up table,and the toxic potency based on the 7-gas model can be obtained.The method was applied into an underground commercial street in Chongqing.It showed that the results between the 7-gas model and 3-gas model(CO,CO 2,and O 2) were obviously different.It indicated that there needs some modifications in conclusions and results from 3-gas model for fire-risk assessments.
A reduced mechanism for propane/air combustion and its flame inhibition by phosphorus-containing compounds (PCCs) is constructed with the level of importance (LOI) method. The analysis is performed on solutions of freely propagating premixed flames with detailed chemical kinetics involving 121 species and 682 reactions proposed by Jayaweera et al. For the non-homogeneous reaction-diffusion system, the chemical lifetime of each species is weighted by its diffusion timescale, and the characteristic flame timescale is used to normalize the chemical lifetime. The definition of sensitivity in LOI is extended so that multi-parameters can be used as sensitivity targets. Propane, oxygen, dimethyl methylphosphonate (DMMP), and flame speed are selected to be perturbed for sensitivity analysis, the species with low LOI index are removed, and reactions involving the redundant species are excluded from the mechanism. A skeletal mechanism is obtained, which consists of 57 species and 268 elementary reactions. Calculations for laminar flame speeds, key flame radicals and catalytic cycles using the skeletal mechanism are in good agreement with those by using the detailed mechanism over a wide range of equivalence ratio undoped and doped with DMMP.