Fuel, Vol.93, No.1, 339-350, 2012
Evaluation of chemical-kinetics models for n-heptane combustion using a multidimensional CFD code
Computational fluid dynamics (CFDs)-based predictions are presented for nonpremixed and partially premixed flames burning vaporized n-heptane fuel. Three state-of-the-art chemical kinetics models are incorporated into a time-dependent, two-dimensional, CFD model known as UNICORN. The first mechanism is the San Diego (SD) mechanism (52 species and 544 reactions), the second one is the Lawrence Livermore National Laboratory (LLNL) mechanism (160 species and 1540 reactions), and the third one is the National Institute of Standards and Technology (NIST) mechanism (197 species and 2926 reactions). Soot model based on acetylene, and radiation model based on optically thin media assumption are included. Two-dimensional calculations are made for the detailed structures of nonpremixed and partially premixed flames, strain-induced extinction and diffusion-controlled autoignition and the results are compared with the available experimental data. Diffusion-controlled autoignition characteristics are also compared with the ignition delay times calculated in homogeneous stoichiometric mixture of n-heptane and air. Through the simulation of complete flowfields between the opposing fuel and air ducts reasons for the flame curvature seen in some experiments are explained. Compared to the traditional one-dimensional models for opposing-jet flames, two-dimensional simulations are found to give results closer to the experimental values when the flames are highly stretched. While LLNL mechanism predicted extinction of a nonpremixed flame better, NIST mechanism predicted the autoignition behavior in the flowfield established by the opposing jets of fuel and heated air better. However, all three mechanisms predicted both the nonpremixed and partially premixed n-heptane flames very well. Surprisingly, SD mechanism with less than one-third of the species used in the other two mechanisms predicted flame structures with nearly the same accuracy. Comparisons made with the available experimental data could not suggest which mechanism is better in predicting the minor species concentrations. Computations could not predict the temperature rise detected in the experiments in the premixed-combustion zone of a partially premixed flame when it was subjected to a moderately high stretch rate. (C) 2011 Elsevier Ltd. All rights reserved.