Data assimilation using high-speed measurements and LES to examine local extinction events in turbulent flames

Data assimilation techniques are investigated to determine how high-speed experimental measurements can be infused into a combustion simulation with the goal of capturing transient combustion events and isolating model deficiencies. To this end, an ensemble Kalman filter (EnKF) is employed to assimilate simultaneous measurements from tomographic PIV and OH-PLIF into a combustion LES of a turbulent DME jet flame, taking into consideration experimental uncertainties and modeling errors. It is shown that by assimilating experimental data, EnKF improves the prediction of the extinction and reignition dynamics observed in this flame. Subsequently, the capability of the assimilation method in evaluating the model performance is examined by considering an assimilation sequence. It is shown that the combustion model investigated (namely a flamelet/progress variable model) exhibits a tendency to relax towards a more reactive state, indicating a deficiency in quantitatively predicting the extent of extinction and reignition with this particular model. (C) 2018 The Combustion Institute. Published by Elsevier Inc. All rights reserved.