Integrated data-model analysis facilitated by an Instrumental Model

A paradigm is described and demonstrated for rigorously evaluating model-versus-data agreement while extracting new insights for improving the model and experiment. “Bound-to-Bound Data Collaboration” (B2B-DC) is augmented with an Instrumental Model, integrating uncertainty quantification of the reactor model, chemical model, and data analysis. The subject of analysis is a fuel-lean C2H2/O2/Ar premixed laminar flat flame, mapped with VUV-photoionization molecular-beam mass spectrometry at the Advanced Light Source of Lawrence Berkeley National Laboratory. Experimental signals were modeled with a CHEMKIN flame code augmented with an Instrumental Model. Consistency of the model and raw experimental data are determined as a quantitative measure of their agreement. Features of the mole-fraction profiles are predicted for O, OH, C2H3, and background contributions to H2O measurements. Also computed are posterior distributions of the initial targets and model parameters, as well as their correlations. This approach to model-versus-data assessment promises to advance the science and practical utility of modeling, establishing validity rigorously while identifying and ranking the impacts of specific model and data uncertainties for model and data improvements.