In the same way we blend ab initio computational chemistry and data science, our group blends experiments and data science. Our experimental goal is to maximize the information generation rate – both increasing the information extractable from data and throughput of data. Our jet-stirred reactor (JSR) facility was designed and built with exactly these ideas in mind. Among other unique features, the facility employs fast-response diagnostics and computer-controllable components to increase throughput and enable automation.
We have been using this facility to gather data at experimental conditions pinpointed by Bayesian Design of Experiments, which identifies the most informative experiments for predictions of a given Quantity of Interest (e.g., ignition delay in a compression ignition engine or NOx emissions in a gas turbine engine).