n-Dodecane Mechanism With ANN-Assisted Reduction for CFD Modeling to Predict Formation of Light-Weight Aromatics and Soot in Diffusion Flames: Comparison With Experimental Data

Abstract

n-Dodecane, a key component in diesel and aviation fuel, is commonly used to simulate real-world diesel and aviation fuels (Jet-A and Chinese RP-3). Since existing n-dodecane kinetic mechanisms may not fully address the complexities of aromatics formation during combustion, this study proposes a mechanism that not only extends the capability of predicting 16 light-weight aromatics but also provides a compact size with improved accuracy in predicting combustion characteristics. Using a two-step reduction method involving path flux analysis (PFA) and artificial neural network (ANN) without tuning kinetic parameters, the newly constructed mechanism consisting of 155 species and 827 reactions is coupled with a 2-D computational fluid dynamics (CFD) model of a laminar diffusion flame that well reproduces experimentally measured centerline profiles of flame temperature, aromatics and soot volume fraction in combustion of methane doped with n-dodecane. From the results obtained by CFD, we investigate the effect of n-dodecane on the spatial distributions of aromatics and reaction pathways, which have not been analyzed in previous literature.