Abstract
The identification of the combustion model is always guided by the compromise between accuracy and computational cost. While species transport models offer accuracy, their computational cost requirement can become prohibitive, especially when de-aling with higher-order hydrocarbon fuels. To mitigate this, the virtual mechanism definition aims to optimize the predictivity minimizing the amount of information to be transported. Thereby the virtual mechanism leverages fictitious species and a few step reactions whose parameters calibration is performed with a genetic algorithm. This work outlines the procedure for the derivation of a virtual reaction mechanism for the study of lean / fuel mixtures with 60% of content (by vol.). These conditions require an adequate characterization of the virtual species differential diffusion oriented to reconstruct the flame sensitivity toward the aerodynamic stretch. After the mechanism derivation, its predictivity has been validated on a swirl-stabilized perfectly premixed turbulent test case. The artificially thickened flame model has been adopted to allow the flame front discretization on an large eddy simulation (LES) grid and to model the turbulence chemistry interaction. The numerical results show a very good agreement with the experimental optical measurements confirming the effectiveness of this approach for predicting the / blend.