| description abstract | During the last decades, several new technologies were investigated in order to reduce the pollutant emissions and increase the overall engine efficiency. Unluckily, some of them including the lean direct injection spray combustion hinder the ignition performances of the combustor. Moreover, several expensive tests under very challenging operating conditions must be carried out to obtain the required certifications and assess the combustor behavior with respect to the ignition process. Therefore, a deeper knowledge of the phenomena involved in the flame onset is mandatory to shorten the design process and achieve the required performances from the very beginning. In the last years, computational fluid dynamics (CFD) simulations established a valid alternative to the experiments to investigate the complex phenomena involved in the ignition process. In fact, several examples are available in scientific literature about the use of simulations to predict the development of the flame starting from an initial kernel. In particular, large eddy simulation (LES) proved to be a reliable tool to uncover new mechanisms of ignition and flame stabilization in gas turbines. In this work, two reactive LES of the ignition process were attempted using ansysfluent 2019R1, with the aim of testing the thickened flame model already implemented in the solver. In fact, compared to the previous versions, a new formulation for the efficiency function based on the pioneering work of Colin was made available. Such promising tool was validated against some detailed experimental results of a lean swirled flame, known as knowledge for ignition, acoustics and instabilities (KIAI)-CORIA spray flame. At first, a non-reactive and reactive LES were carried out to validate the cold field and the stabilized flame structure respectively. Finally, two ignition simulations were performed, from initial spark deposition up to flame stabilization or kernel quenching. All the obtained results have been extensively compared against the available experimental data showing that the employed simulation setup is fairly capable of describing the phenomena involved in the rig ignition. | |
