Combining Multiscale Physics and Data to Enable Predictive Modeling of Complex Reacting Systems in Energy, Propulsion, and the Environment

Understanding and being able to predict the outcomes of complex reaction systems would be invaluable to varied scientific and engineering disciplines – ranging from enabling future energy, aerospace, and chemical processing technologies to understanding the Earth’s climate and the potential origins of life beyond Earth.  Yet, unraveling and making predictions of complex reaction systems are notoriously challenging tasks.  I outline an overall approach that overcomes those challenges by combining (1) ab initio computational studies of historically missing physics and chemistry, (2) data-driven modeling that integrates data and physics from multiple scales, and (3) optimal design of experiments and theoretical calculations. I will then highlight one of these three elements in depth, focusing on our ab initio theoretical studies of non-equilibrium phenomena induced by reactions in combustion and propulsion and by photons in the Earth’s atmosphere – both of which feature vibrationally excited reactants relevant to “vibrational ladder climbing” in plasmas.  Finally, I will describe how these theoretical calculations, in general, can be combined with experimental data via multiscale data-driven modeling and, more broadly, how automation of our overall approach can create “robotic scientific communities” capable of autonomous scientific inquiry and self-improving simulations of reacting systems.