The combustion of fuels, whether they are fossil or synthetic, is and will remain an important source for the primary energy production. In order to support the design process of novel combustion devices engineers need reliable and efficient simulation tools. The flames in industrial combustors are usually affected by turbulence and lots of applications operate in a partially premixed mode, including extinction and re-ignition effects and mixed flame modes. These fame configurations are difficult to model. In the past, it has been shown that especially the Flamelet/Progress variable (FPV) approach and the Eulerian Stochastic fields (ESF) method, which belongs to the family of transported PDF methods, are suitable to model partially premixed flames. The FPV approach is computationally relatively cheap, as it uses tabulated chemistry and a presumed sub-grid probability density function (PDF). On the contrary, the ESF method is assumed to be more accurate, but also more expensive, as it uses finite-rate chemistry and directly solves the sub-grid PDF with additional stochastic transport equations. The objective of this work is to analyze and quantitatively compare the ESF method with the FPV approach in simulations of turbulent partially premixed flames within the context of time resolving Large Eddy Simulations (LES). The focus is on the ESF method and it is investigated how the choice of the specific ESF formulation, the chemical mechanism, the number of stochastic fields, the turbulence model, and the computational grid affect the overall prediction quality of the simulation. These results are then directly compared to simulations conducted with the FPV approach using both premixed and non-premixed manifolds. In addition, a quantitative analysis is carried out using the Wasserstein metric to measure the dissimilarity between experimental and numerical data sets. In the present work, three relatively new turbulent methane/air flame configurations that show partial premixing are investigated. The worst case is a flame with inhomogeneous fuel inlet, which exhibits a mixed flame mode and shows a moderate degree of extinction. The second flame is a piloted non-premixed diffusion flame that is close to blow-off and shows a strong degree of local extinction. Finally, an oxy-fuel jet flame is investigated, which has a high proportion of hydrogen in the fuel stream. In order to model differential diffusion effects properly, the ESF method has been extended in this work to account for detailed molecular transport. The analysis suggests that the ESF method should be the preferred combustion model in the simulation of all three flame configurations, in order to obtain the most accurate results for all analyzed quantities of interest. It is shown, however, that the method is relatively insensitive to the selected number of stochastic fields and that other parameters, such as the choice of the chemical mechanism are more relevant for the accuracy of the results. This indicates that in LES the sub-grid turbulence chemistry interaction may be deliberately neglected, given a finite rate mechanism is used. In addition, it is shown that the FPV approach also provides satisfying results for some quantities of interest, which are compatible to the ESF results. Finally, the simulations of the oxy-fuel fame indicate that, given the high content of hydrogen, differential diffusion effects need to be incorporated to obtain reasonable results.    «

The combustion of fuels, whether they are fossil or synthetic, is and will remain an important source for the primary energy production. In order to support the design process of novel combustion devices engineers need reliable and efficient simulation tools. The flames in industrial combustors are usually affected by turbulence and lots of applications operate in a partially premixed mode, including extinction and re-ignition effects and mixed flame modes. These fame configurations are difficul...    »