Influence of nonthermal termolecular reactions on the optimization and uncertainty quantification of a hydrogen combustion model

Nonthermal termolecular reactions were shown to have significant impacts on combustion predictions of H 2 , CH 4 , and other fuels. But they are not considered in earlier model development effort s, many of which involve tuning or optimization. While adding new chemistry to optimized models is not recommended, we need to quantify its influence. The impact on model optimization has been examined on the H 2 sub-mechanism of the Foundational Fuel Chemistry Model Version 1.0. The current work compared the optimization using the model as is, and with termolecular chemistry fully implemented, with new termolecular reactions and corrections to the stabilization reactions. The models optimized with and without termolecular chemistry have comparable performance and both could reconcile all the experimental targets considered. The two models show similar predictions under most target conditions, which can be explained by the compensation of kinetic parameters within their uncertainty ranges. Extrapolation of the model to conditions outside the target list shows larger discrepancy. While models with proper inclusion of nonthermal reactions should in general give more reliable predictions, it is expected that the difference can be minimized with a proper target list and optimized models can be used with attention on conditions very sensitive to termolecular chemistry. The conclusions provided guidance on model optimization with termolecular chemistry in consideration. (c) 2023 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Automated summary

Nonthermal termolecular reactions have significant impacts on combustion predictions, but have not been considered in earlier model developments. This study compared model optimization with and without termolecular chemistry, and found comparable performance. Attention should be paid to sensitive conditions in the application of the models.