Numerical and Experimental study of aircraft engine ignition

The aim of this study is to develop and validate an ignition model, which could be coupled with a CFD code to predict ignition/re-ignition of an aircraft engine. This model is validated using a data base acquired on the MERCATO test rig (ONERA/Fauga-Mauzac), reproducing a simplified combustion chamber. A simulation of the two-phase flow inside the chamber has been performed and compared with experimental data for a non-reactive flow, just before ignition. The spray ignition model has been tested for several positions of the spark plug device. Introduction With the reduction of pollutant emissions level, industrials must design novating combustion chamber, allowing lean combustion. In the case of relight at high altitude, corresponding to critical conditions, ignition can’t be correctly predicted by current methods used in industry. A parametric experimental study on a real combustion chamber is very expensive, and cannot be used to test all possible geometries during development process. CFD remains the cheaper solution, although modelling of the spray ignition is rather complex. Moreover, the complete modelling of spark discharge recquires great computational cost, to take into account plasma effects or detailed chemistry schemes. In the litterature, there are few studies about two-phase flow combustion for industrial configurations. Widmann [1] worked on a methanol spray in a swirled air flow, characterizing the droplet velocity with and without combustion. Ikeda et al [2] studied the influence of pressure on a burning diesel spray in a highly-pressurized swirlstabilized combustor. Hochgreb et al [3] studied relight at high-altitude conditions for a lean combustor. With high speed imaging, they tracked the motion and break-up of the flame during the propagation phase. Recently, Mastorakos et al [4] studied the ignition of a n-heptane spray with multiple sparks. Using a mobile spark plug device, they established a map of ignition probability into a closed vessel, and identified parameters optimizing ignition. Through three thesis, ONERA developed a simplified spray ignition model. This model can be coupled with a RANS code to simulate ignition of a combustion chamber by a spark plug. Quintilla[5] has developed a 0D ignition model, allowing to compute the temperature evolution of the ignition kernel. Ouarti[6] used this model to perform a 2D axisymmetric RANS simulation of several ignition cases. Although these simplifications, ignition tendencies were well reproduced, and the model allowed to discriminate successful ignition cases. Recently, García-Rosa[7] developed an 1D ignition model, which computes the growth of the ignition kernel. This model has been partially validated for gaseous, monodisperse and polydisperse two phase mixtures[8]. In the present work, the 1D ignition model is tested with the results of a 3D simulation of a two-phase flow on the MERCATO test rig. Non-reactive simulations are performed using the CFD code CEDRE (ONERA), with RANS method for the air flow and Lagrangian stationnary approach for the liquid flow. Results are validated with LDA (Laser Doppler Anemometry) and PDI (Phase Doppler Interferometry) measurements. For the liquid phase, stationnary approach allow to carry out fast simulations, but cannot be used to perform a complete ignition simulation of a chamber, because of the unsteady nature of the phenomena. Until now, we limit our study to the early formation of the ignition kernel. The kernel transport is out of the scope of our work. Spray kernel ignition model The aim of the present model is to simulate the early propagation of an ignition kernel in a two-phase mixture. The energy deposition is considered as an instantaneous, adiabatic and isobaric heating process, which creates a spherical flame. The expansion or extinction of this kernel depends on the local equivalence ratio. With these hypothesis, the kernel growth phase is described by numerical resolution of conservation equations for 1D dilutespray mixture (See Eqs. 1, 2 and 3). ∗Corresponding author: guillaume.linassier@onera.fr