Detailed kinetic models help us investigate alternative fuels for the next generation of cleaner combustion devices. While automated mechanism generators like Reaction Mechanism Generator (RMG) can build these models, the first results are usually not accurate enough to use without modification. We are developing a highly automated workflow that can iteratively improve these models without relying on experimental data.
The workflow analyses thousands of model parameters to determine which ones contribute the most to the overall error.
Then the workflow improves the top ten parameters by calculating them with quantum chemistry software, and feeds the results back into RMG for the next round of improvement.
