Assessment of a model for emission subgrid-scale turbulence-radiation interaction applied to a scaled Sandia flame DD.

• A subgrid-scale (SGS) turbulence-radiation interaction model's accuracy is assessed. • The configuration under study is a scaled version of the Sandia flame D. • The reference solution is given by the Eulerian Monte Carlo stochastic field method. • The model performs slightly better than an approach that neglects SGS fluctuations. • A major inaccuracy of the model is in its estimate of the SGS temperature variance. A simple model to account for subgrid-scale turbulence-radiation interaction (SGS-TRI) in the filtered radiative emission is evaluated for coupled and uncoupled simulations of a scaled-up version of the Sandia flame D. The model is assessed based on comparisons with a reference solution that accounts for SGS-TRI by employing the Eulerian Monte Carlo stochastic field method (ESF). While SGS-TRI effects are in general marginal for the studied flame, results of the uncoupled calculations show that the model consistently gives a better approximation of the filtered emission than what is obtained by fully neglecting SGS fluctuations. However, this improvement is small, due to a significant underprediction of the SGS temperature variance T ″ 2 ˜ , which was determined by assuming local equilibrium between the dissipation and the production of this quantity. If a more accurate estimate of T ″ 2 ˜ is adopted (such as the one computed with the ESF method), the studied model is able to qualitatively capture the contribution of the temperature auto-correlation to the filtered emission, and its performance can likely be enhanced by choosing an adequate value for its parametric constant. When applied to the coupled simulation, the accuracy of the model, measured in terms of its results for the filtered temperature, species concentration, radiative emission and radiative heat source, is at best on par with the approach that completely neglects SGS-TRI. This is true even if the SGS temperature variance obtained from the ESF method is used, likely because the value of the constant associated to the model is too large for the flame under consideration. [ABSTRACT FROM AUTHOR]