Introduction of auto-ignition in the Thickened Flame model for Large Eddy Simulations of reheat systems

An extension of the Dynamic Thickened Flame model (DTF) Legier et al., (2000) called DTF_AI (Dynamic Thickened Flame model for Auto-Ignition) meant to provide a more accurate prediction of auto-ignition delay 1 AI in Large Eddy Simulations (LES) of industrial reheat systems is proposed. A methodology to estimate the quantities required by the new combustion model ie. the local laminar flame thickness delta(0)(l), laminar flame speed S-l(0), and progress variable c is at first presented. Then, a modification of the classical DTF model consisting in a modulation of the thickening factor as a function of the progress variable gradient (del(c)) is introduced to improve the prediction of auto-ignition events on coarse LES grids. A comparison between DNS and LES of an academic test case featuring a partially premixed two-phase flow representative of aircraft engines afterburners shows that the new combustion model gives satisfying results for both propagation and auto-ignited flame fronts encountered in real reheat systems. While the classical DTF model induces significant errors for auto-igniting cases, the new model predicts the flame location with reasonable accuracy in 3D, partially premixed, unsteady two-phase flow representative of reheat burners.