Ensemble Kalman Filter for Assimilating Experimental Data into Large-Eddy Simulations of Turbulent Flows

Data assimilation techniques are investigated for integrating high-speed high-resolution experimental data into large-eddy simulations. To this end, an ensemble Kalman filter is employed to assimilate velocity measurements of a turbulent jet at a Reynolds number of 13,500 into simulations. The goal of the current work is to examine the behavior of the assimilation algorithm for state estimation of turbulent flows that are of relevance to engineering applications. This is accomplished by investigating the impact that localization, measurement uncertainties, assimilation frequency, data sparsity and ensemble size have on the estimated state vector. For the flow configuration and computational setup considered in this study an optimal value of the localization radius is identified, which minimizes the error between experimental data and state vector. The impact of experimental uncertainties on the state estimation is demonstrated to provide solution bounds on the assimilation algorithm. It is found that increasing the number of ensembles has a positive impact on the state estimation. In comparison, decreasing the assimilation frequency or reducing the experimental data available for assimilation is found to have a negative impact on the state estimation. These findings demonstrate the viability of assimilating measurements into numerical simulations to improve state estimates, to support parameter evaluations and to guide model assessments.