Combustion kinetic model optimization using the derived targets from MBMS experiments

Abstract

Synchrotron-based molecular-beam mass spectrometry (MBMS) measurements can yield the concentrations of various species during combustion experiments. These quantitative data are helpful for combustion kinetic model development, validation, and optimization. The uncertainties of stable species can be limited within 30%, but for free radicals, the uncertainties can be large with a factor of 2–4 owing to the unknown photoionization cross sections. Therefore, measurements of free radicals are often used qualitatively for model validation, and how to utilize these measurements quantitatively for model optimization is still a challenge. In the present work, we propose to use derived targets, which are the ratios of the mole fractions of the same free radicals under two different experimental conditions, to optimize kinetic models. These derived targets have much smaller uncertainties compared to the original mole fractions because the unknown photoionization cross sections of the free radicals are canceled out, leading to more effective optimization results. To demonstrate this method, a dimethyl ether (DME) system and a methanol system are chosen as case studies. A global-sensitivity-based model analysis is first conducted to select informative targets for model optimization, i.e., derived targets of C2H5 in the DME system and the derived targets of CH3 in the methanol system. Model optimization is then performed using the ANN-MCMC method with these selected derived targets. The results of the global sensitivity analysis show that some important reactions can be highlighted when the derived targets are considered, while these reactions may not be important when considering only the original targets. The results of the comparison of two kinds of the optimized models show that better posterior models can be achieved by the model optimization based on derived targets than based on the original targets.