(659c) Heterogeneous Catalysis Kinetic Characterization Via Sparse Graphs

Kinetic rate estimation is essential to understanding the underlying elementary behavior of a catalyst. While computational kinetic methods are necessary for estimating an unknown system, there may be drawbacks in sensitivity to the selection of the reaction model, estimates of parameter conditions and inclusion of experimental results. Combining the use of physical models and experimental transient data, accurate kinetic rate information can be obtained with no assumption on the reaction mechanism while reducing the assumptions required on the experiment. We leverage data science and statistical techniques in the form of Gaussian Graphical Models to extend existing methodology in predicting kinetic rates. This methodology is applied to highly dense transient data from the Temporal Analysis of Products (TAP) reactor allowing for determination of the kinetics in an intrinsic manner within time as well as between experiments. The analysis results in a series of connected graphs to make better insights of the catalyst behavior as the surface evolves.