The application of porous media to simulate the upstream effects of gas turbine injector swirl vanes

Numerical simulations are an invaluable means of evaluating design solutions. This is especially true in the initial design phase of a project where several simulations may be required as part of an optimisation study. The design of aircraft gas turbine combustor external aerodynamics frequently calls upon the services of numerical methods to visualise the existing flow field, and develop architectures which improve the performance of the system. Many of these performance improvements are driven by the desire to reduce fuel burn and cut emissions lowering the environmental impact of aviation. The gas turbine combustion chamber is, however, reasonably complex geometrically and requires a high fidelity model to resolve small geometric details. The fuel injector is the most geometrically complex component, requiring around 20% of the mesh cells of the entire domain. This makes it expensive to model in terms of both requisite computational resource and run time. Most modern aircraft gas turbines utilise swirling flow fields to stabilise the flame front in the combustion liner. The swirl cone is generally generated using fixed angle vane rows within the injector. It is these small features that are responsible for the requisite high mesh cell count. This paper presents a numerical method for replacing the injector swirl vane passages with mathematically porous volumes which replicate the required pressure drop. Modelling using porous media is preferential to modelling the fully featured injector as it allows a significant reduction in the size of the computational domain and number of cells. Additionally the simplification makes the geometry easier to change, scale and re-mesh during development. This in turn allows significant time savings which serve ultimately to expedite the design process. This method has been rigorously tested through a range of approach conditions and flow conditions to ensure that it is robust enough for use in the design process. The loss in accuracy owing to the simplification has been demonstrated to be less than 4.4%, for all tested flow fields. This error is dependent on the flow conditions and is generally much less for passages fed with representative levels of upstream distortion.