The use of lean burning flames stabilized by highly swirling flows represents the most effective technology to limit NOx emissions in modern aeroengine combustors. In these devices up to 70% of compressed air is admitted in the combustor through the injection system, which is usually designed to give strong swirling components to air flow. Complex fluidynamics is observed with large flow recirculations due to vortex breakdown and precessing vortex core, that may result in a not trivial interaction with liner cooling flows close to combustor walls. This interaction and its effects on the local cooling performance make the design of the cooling systems very challenging and time-consuming, considering design and commission of new test rigs for detailed analysis. Keeping in mind costs and complexities related to the investigation of swirl flow/wall cooling interaction by experimental approach, CFD can be considered an accurate and reliable alternative to understand the associated phenomena. The widely known overcomes of RANS formulation (e.g. underestimation of mixing and inability to properly describe swirling flows) and the more and more impressive increase in computational resources, pushed hybrid RANS-LES models as valuable and affordable approaches to accurately solve the main turbulent flow structures.
This work describes the main findings of a CFD analysis intended to accurately investigate the flow field and wall heat transfer as a result of the mutual interaction between a highly swirling flow generated by a lean burn nozzle and a slot-effusion liner cooling system. In order to overcome some limitations of RANS approach, the simulations were performed with SST-SAS, a hybrid RANS-LES model. Moreover, the significant computational effort due to the presence of more than 600 effusion holes was limited exploiting two different modelling strategies: a homogeneous model based on the application of uniform boundary conditions on both aspiration and injection sides, and another solution that provides a coolant injection through point mass sources within a single cell. CFD findings were compared to experimental results coming from an investigation carried out on a three sector linear rig. The comparison pointed out that advanced modelling strategies, i.e. based on discrete mass sources, are able to reproduce the effects of mainstream-coolant interactions on convective heat loads. Validated the approach through a benchmark against time-averaged quantities, the transient data acquired were examined in order to better understand the unsteady behaviour of the thermal load through a statistical analysis, providing useful information with a design perspective.