On the modelling of liquid fuel ignition and atomization in aero engine combustors

A fast and reliable relight of aero engines burners is one of the most critical points to ensure aircraft safety. The so-called altitude relight is the process that allows the combustor to be re-ignited after a flame-out during flight. Several expensive tests must be carried out to obtain the required certifications, which makes important to fully understand the problem of the flame onset. To speedup the design process, Computational Fluid Dynamics established as valid alternative to the experiments to investigate the complex phenomena involved in the ignition process. In this work, a fully reactive Large Eddy Simulation of the ignition process is attempted with the aim of validating the setup for future applications within an advanced design process. In this line, a throughout validation is carried out against some detailed experimental results of a lean spray flame. At first, a non-reactive and reactive simulations are carried out to validate the cold flow field and the stabilized flame structure. Then, an ignition simulation is performed, from initial spark deposition up to flame stabilization. The obtained results are extensively compared with the available experimental data, showing that the employed simulation setup can describe quite well the phenomena involved in the rig ignition. However, this procedure does not allow to perform statistical studies, such as the optimization of the igniter position. This issue is critical since it can help to minimize the amount of energy required for ignition and to increase the durability of the hardware. In fact, several spark discharges must be simulated for each position, to account for different realizations due to the turbulent flow field. Therefore, the previous Large Eddy Simulation approach is not feasible in this scenario, due to the excessive computational effort. In scientific literature, specific low-order models were developed to provide an affordable estimation of the local ignition probability. Due to the large amount of assumptions they clearly sacrifice part of the accuracy and of the physical consistency, but the short turnaround time makes them the first choice at low Technology Readiness Level. In this thesis, a low-order design model is implemented and used to investigate the ignition probability of the same test rig simulated in detail, showing that it can provide good results if a careful sensitivity study is firstly conducted. This set of simulations represents a first attempt in the scientific literature to carefully validate a low-order model using a flow-field from LES and very accurate experimental data. Moreover, the LES of the ignition process is used to further validate the model and highlight its main shortcomings. To the best of the author knowledge, such in-depth validation was not attempted so far. The last step of the thesis concerns the development of a simulation and post-processing methodology to investigate the primary breakup of the spray in altitude relight conditions. Such activity is motivated by the fundamental importance of spray initialization in full chamber simulations. In fact, due to the very poor atomization quality and the negligible evaporation process, the droplets ejected from the nozzle can travel all along the combustion chamber and finally reach the combustor walls. Under these conditions, the ignition tools already described might fail if the spray is injected with a wrong distribution. Therefore, a tentative approach to evaluate the spray size distribution is proposed in the last section of this work. It is based on a run-time evaluation of distributed variables such as liquid/gas density of interface or locally defined variables as the curvature of the interface. In the present work, the Eulerian Lagrangian Spray Atomization model is used as a basis to test such technique. An academic planar prefilmer configuration is selected to evaluate the accuracy of the post-processing, thanks to the availability of experimental measurements in the proximity of the injector. The reported results include a brief description of the simulated liquid structures, the predicted Sauter Mean Diameter, the evolution of interface curvature and a final proposal to derive the spray size distribution. The novelty of this last section is mainly represented by the use of the interface curvature to analyze and postprocess the primary breakup. At the best of the author knowledge, this is the first time that such an approach is attempted to an actual atomizing device. Overall, the aim of this work is to validate and to develop a set of tools to improve altitude relight design in aero engines. Although much work is still needed, this thesis represent a first step towards the use of more advanced numerical tools to optimize this process and to more easily meet the required certifications.