Comparison of different DRG-based methods for the skeletal reduction of JP-8 surrogate mechanisms

Directed relation graph (DRG) techniques are used to generate small skeletal mechanisms capable of accurately simulating the combustion of a two-component surrogate for JP-8 jet fuel. Within the DRG framework, six different reduction techniques are considered, and the effectiveness of different definitions for the connection weights and error propagation is evaluated. The use of DRG reduction techniques for aided sensitivity analysis (DRGASA) and for on-the-fly reduction (flux-based DRG) is studied in detail. An optimal reduction approach based on the sequential use of DRG, DRGASA, and flux-based DRG is proposed. When global reduction is applied to a detailed mechanism of 234 species and 6997 reactions, the six reduction techniques result in very different skeletal mechanisms, but all of them are essentially equivalent in terms of accuracy and number of retained species (82–92 species). Finally, for two-dimensional coflow flame test problems, on-the-fly DRG techniques are investigated. Error-propagation-based methods are found to extinguish the flame artificially and cannot be used in an on-the-fly implementation. Conversely, normal DRG methods greatly improve the mechanism reduction, and accurate solutions are obtained using about 15% of the detailed mechanism.