A data-driven kinematic model of a ducted premixed flame

Reduced-order models of flame dynamics can be used to predict and mitigate the emergence thermoacoustic oscillations in design gas turbine rocket engines. This process is hindered by fact that these models, although often qualitatively correct, are not usually quantitatively accurate. As automated experiments numerical simulations produce ever-increasing quantities data, question arises as how this data assimilated into physics-informed reduced-order order render In study, we develop test a physics-based model ducted premixed which parameters learned from high-speed videos flame. The experimental level-set solver using an ensemble Kalman filter. leads optimally calibrated with quantified uncertainties, accurately reproduces elaborate nonlinear features such cusp formation pinch-off. continues match after assimilation has been switched off. Further, model, extracted automatically, shown first-order behavior expected on physical grounds. study shows updated rapidly whenever new or becomes available, without itself having stored.