An efficient method to reproduce the effects of acoustic forcing on gas turbine fuel injectors in incompressible simulations

Previous studies have highlighted the importance of both air mass flow rate and swirl fluctuations on the unsteady heat release of a swirl stabilised gas turbine combustor. The ability of a simulation to correctly resolve the heat release fluctuations or the flame transfer function (FTF), important for thermoacoustic analysis, is therefore dependent on the ability of the method to correctly include both the swirl number and mass flow rate fluctuations which emerge from the multiple air passages of a typical lean-burn fuel injector. The fuel injector used in this study is industry representative and has a much more complicated geometry than typical premixed, lab-scale burners and the interaction between each flow passage must be captured correctly. This paper compares compressible, acoustically forced, CFD (computational fluid dynamics) simulations with incompressible, mass flow rate forced simulations. Incompressible mass flow rate forcing of the injector, which is an attractive method due to larger timesteps, reduced computational cost and flexibility of choice of combustion model, is shown to be incapable of reproducing the swirl and mass flow fluctuations of the air passages given by the compressible simulation as well as the downstream flow development. This would have significant consequences for any FTF calculated by this method. However, accurate incompressible simulations are shown to be possible through use of a truncated domain with appropriate boundary conditions using data extracted from a donor compressible simulation. A new model is introduced based on the Proper Orthogonal Decomposition and Fourier Series (PODFS) that alleviates several weaknesses of the strong recycling method. The simulation using this method is seen to be significantly computationally cheaper than the compressible simulations. This suggests a methodology where a non-reacting compressible simulation is used to generate PODFS based boundary conditions which can be used in cheaper incompressible reacting FTF calculations. In an industrial context, this improved computational efficiency allows for greater exploration of the design space and improved combustor design.