Modelling of the Flow Field Within a Generic Aero-Engine Combustor

Abstract

A generic combustor has been modelled using Reynolds Averaged Numerical Simulations (RANS) and Large Eddy Simulations (LES). The combustor is representative of a sector of an aero-engine combustor, including one fuel injector and several dilutions ports. To enable optical measurements the combustor has been equipped with air-cooled quartz windows. Both RANS and LES computations have been performed, and two grids have been applied. One grid represents the combustor only, while the other also includes the burner swirler passages. The difference between applying the boundary conditions at combustor inlet and including the swirlers in the CFD model are studied. The computations are performed with a Rolls-Royce in-house code. The code is block-structured and parallelised using MPI (Message Passing Interface). Both isothermal and combustion computations have been carried out. The combustion process has been modelled with a conserved scalar flamelet model, but also a finite rate chemical model has been applied. Detailed experimental data are available (LDA, PIV, DGV, Raman, CARS) with the primary and dilution zone of the combustor. The measurements have been performed by DLR Cologne within the EC funded research project MOLECULES. The cold flow field in the primary zone is represented well by both the RANS and LES model, but if the injector is modelled, the combustor inlet profiles are better represented by the LES model. In the dilution zone there are larger differences between LES and RANS. The RANS computations underestimate the jet penetration. The combusting simulations using the conserved scalar model compare not very well with the experimental results close to the injector: the mixture burns too fast. The finite rate reaction model, combined with RANS gave better results. It is concluded that in the vicinity of the injector the chemistry cannot be assumed to be infinitely fast.