Automatically generated detailed and lumped reaction mechanisms for low- and high-temperature oxidation of alkanes

In this work, we present a methodology on automatic generation of predictive lumped sub-mechanisms for normal and branched alkanes. This methodology aims at obtaining lumped reaction mechanisms that preserve the chemical behavior of each reaction class in the detailed model. To achieve this goal, detailed sub-mechanisms for combustion of alkanes are generated by employing an updated version of the MAMOX ++ software developed in this work; recent progress in the low-temperature reaction classes and rate rules are incorporated into the updated software. Instead of computing the selectivities of several primary products with MAMOX ++ and fitting the selectivities between the detailed and lumped models, this work proposes a new methodology to generate the lumped sub-mechanisms for fuel molecules. The stoichiometric parameters and the reaction rates for each reaction class in the lumped sub-mechanism are fitted to match those in the detailed model. Based on the present methodology, both the detailed and lumped sub-mechanisms for normal C5-C10 alkanes and branched C5-C8 alkanes, that is for 15 different fuels, are automatically generated and merged into a base chemistry model (i.e. AramcoMech 2.0), respectively. The detailed and lumped models are validated against the experimental data in the literature. The automatically generated detailed models for alkanes are able to capture the experimental targets across a wide range of conditions, demonstrating the robustness of the reaction classes and rate rules adopted. The lumped models for normal alkanes have similar performance to their respective detailed models, and are able to predict the oxidation behavior of normal alkanes. However, prediction deviations between the detailed and lumped models for branched alkanes are shown to be slightly greater.& COPY; 2022 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Automated summary

This work presents a methodology for automatically generating predictive lumped sub-mechanisms for normal and branched alkanes. Detailed sub-mechanisms for alkanes are generated using an updated version of the MAMOX ++ software, incorporating recent progress in low-temperature reaction classes and rate rules. The methodology proposes a new approach to generate lumped sub-mechanisms for fuel molecules by fitting stoichiometric parameters and reaction rates to match those in the detailed model. Validations show that the detailed models capture experimental targets well, while the lumped models perform similarly for normal alkanes but have slightly greater deviations for branched alkanes.