CAREER: Extrapolatable, Uncertainty-Quantified Modeling of Nitrogen Kinetics Informed by Data Across Multiple Scales award

Predictive computer models have immense potential to enable faster, cheaper design of cleaner, more efficient engines that our society and planet urgently need. To have the highest impact on designing engines, models must make accurate predictions with known uncertainty, especially for new cutting-edge engine designs never manufactured or tested. Such predictive models would be especially useful for minimizing the formation of nitrogen oxides (NOx), which are responsible for smog, ground-level ozone, and other effects detrimental to human and environmental health. The goal of this project is to create and validate a predictive model for NOx formation during combustion. An innovative approach, which leverages modern data science and computational chemistry, will be used to create predictive models with known uncertainty. The resulting methodology and models will then form the backbone for future studies of NOx formation during combustion of all conventional and alternative fuels. The project will also engage local high-school students in chemistry and data-science projects, create and disseminate lesson plans for high-school and university teachers, and partner with industry to enable the research to lead to better engine designs immediately.

The technical objective is to create an extrapolatable, uncertainty-quantified, foundational NOx kinetic model by optimally selecting, creating, and exploiting data from molecular to macroscopic scales. The approach fuses (1) theoretical calculations to create molecular data and develop rate laws and mixture rules to represent the pressure and composition dependence of reaction rates, (2) experimental measurements to gather macroscopic data at conditions that best inform engine predictions, and (3) uncertainty-quantified modeling based on multiscale data to create models trustable for predictive design. This work largely centers on undiscovered pathways hypothesized to comprise a major NOx route at the high pressures and low peak temperatures of high-efficiency, low-NOx engines. Altogether, the research will address key outstanding issues in the present understanding of NOx formation at high pressures and produce the first uncertainty-quantified NOx kinetic model constrained by multiscale data. More broadly, the present rate laws, mixture rules, and multiscale data-driven approach will also enable better models of, simulation codes for, and understanding of many other chemically reacting gases.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.