Abstract
Real fuels are complex mixtures of hundreds of molecules, which makes it challenging to unravel their combustion chemistry. Several approaches in the literature have helped to clarify fuel combustion, including multi-component surrogates, lumped fuel chemistry modeling, and functional-group based methods. This work presents an innovative advancement to the lumped fuel chemistry modeling approach, using functional groups for mechanism development (FGMech). Stoichiometric parameters of lumped fuel decomposition reactions dictate the population of the key pyrolysis products, previously obtained by fitting experimental data of real-fuel pyrolysis. In this work, a functional group-based approach is proposed, which can account for real-fuel variability and predict stoichiometric parameters without experimentation. A database of the stoichiometric parameters and/or yields of key pyrolysis products was first constructed for approximately 50 neat fuels, based on previous pyrolysis data and a lumped kinetic model we developed. The effects of functional groups on the stoichiometric parameters and/or yields of key pyrolysis products were then identified and quantified. A quantitative structure-stoichiometry relationship was developed by multiple linear regression (MLR) model, which was used to predict the stoichiometric parameters and/or yields of key pyrolysis products based on ten input features (eight functional groups, molecular weight, and branching index). Products from the pyrolysis of surrogate mixtures and real-fuels were predicted using the MLR model and validated against experimental data in the literature. Comparison with the stoichiometric parameters from the HyChem experiment-based approach (Xu et al., 2018) showed that the predicted values in this work were in reasonable agreement (generally within a factor of two). When the stoichiometric parameters in the jet fuel (POSF 10325) HyChem kinetic model were replaced with this functional-group based prediction, only minor discrepancies were observed in the predictions of key pyrolysis products and global combustion parameters (such as ignition delay times and laminar flame speeds). Sensitivity analysis on stoichiometric parameters revealed their different roles in predicting speciation and global parameters. The functional group approach for predicting stoichiometric parameters in this work was the first step towards developing FGMech for modeling real-fuel combustion chemistry. Further development of the FGMech model's thermodynamic, kinetic, and transport data will be presented in a following study.