APPLICATION OF AN IMPROVED WORKFLOW FOR THE IDENTIFICATION OF FLAME DYNAMICS TO SWIRL STABILIZED WET COMBUSTION

The estimation of a flame transfer function (FTF) from time series data generated by large eddy simulation (LES) via system identification (SI) is an established and important element of thermoacoustic analysis. To achieve results with low uncertainties, continuous (non-truncated) time series of adequate length are required. This can be a challenge in the case of turbulent flames, where signal-to-noise ratio may be low. Because LES codes do not scale to an arbitrarily high degree of parallelism, the wall-clock-time required for confident FTF estimation may be excessive. The present paper tackles this challenge by exploiting characteristic features of linear time-invariant systems. Specifically, we explore how the superposition of multiple simulations with the same excitation signal, but varying initial conditions, increases signal-to-noise ratio and leads to more robust identification, i.e. estimates with narrow confidence intervals. In addition, the established SI approach, which relies on broadband excitation, is compared to excitation with approximate Dirac and Heaviside signals, with the advantage of simpler pre- and postprocessing. Results demonstrate that the proposed workflow, which is inherently parallel because it is based on superposition, reduces significantly the wall-clock-time required for robust FTF identification in the case of broadband excitation. This reduction in wall-clock-time requires more simultaneous simulation runs, i.e. more parallel computational resources, but it does not significantly increase the overall computational cost. Furthermore, the new workflow facilitates FTF estimation when exciting with a Heaviside signal. The proposed method is assessed for the case of a partially premixed, steam enriched ("WET") swirl burner with significant turbulent noise levels. Steam enrichment is a combustion concept that reduces harmful emissions such as NO_x and CO₂ while increasing engine efficiency. However, the effect of steam on the flame response, i.e. the FTF, needs to be better understood. To this end, a combustion model that includes an optimized global chemical mechanism for partially premixed wet methane combustion, has been formulated and validated against experimental data.