Performances of approximate radiative property models for conditions met in ammonia combustion

Abstract

Recent investigations on ammonia as a sustainable fuel highlight the need for proper modelling of the radiative heat transfer in ammonia–air flames. As a result, the present work reports on spectral databases that have been specifically generated, leading to the development of various spectral models, ranging from narrow band models to global ones. These models have been tested against the exact line-by-line (LBL) model for an extended pressure range (1–50 atm) in both premixed and non-premixed flame configurations. Noticeably, the premixed flame structure, which can be represented by a two-layer system, is a challenging configuration for correlated-k distribution models. For this configuration, a detailed investigation of the spectral models on various flame fields led to the conclusion that in spite of a higher computational cost, the Statistical Narrow Band Correlated k-distribution (SNBCK) models are better suited for modelling the radiative processes compared to the full spectrum models, such as the Rank Correlated Full-Spectrum k-distribution (RCFSK), Multi-Scale Rank Correlated Full-Spectrum k-distribution (MSRCFSK), and Weighted Sum of Grey Gases (WSGG). Considering the computational load, a reduced SNBCK model, referred to as SNBCK25, was developed and demonstrated high agreement with the LBL model. As an illustration, for a spherically expanding premixed flame at 20 atm, the SNBCK model had a mean absolute relative error (MARE) of 1.17%, while the SNBCK25, RCFSK with double integration, MSRCFSK, and WSGG models resulted in MARE of 1.87%, 4.62%, 8.23%, and 34.7%, respectively, against the LBL model. It is worth mentioning that the contributions of $\mathrm{NO}$, ${\mathrm{NO}}_2$, and $N_{2}O$ were not significant to the evaluation of the divergence of radiative heat flux in the LBL model. Hence, these three species were ignored in the subsequent radiative computations.