When trying to reproduce the dynamic response of real aero-engines it is important to consider the injection of liquid fuel. The extension of these studies to thermoacoustic instabilities increases the importance of correctly reproducing the liquid droplet behavior (i.e. their dynamics, polydispersity and evaporation), since the feedback loop between the acoustic field and the flame is coupled to the particle dynamics. The interaction between liquid particles and internal element of the combustor is another key aspect. Indeed, in many aero-engine injection systems called airblast atomizers, the liquid fuel is injected from a high-pressure atomizer on a metal surface called prefilmer with the aim of better atomizing the liquid fuel coming from the pressurized tank. The physics of this process is indeed quite complex given the current incapacity to experimentally visualize these regions in real applications or in experimental setups with or without thermoacoustic instabilities. The importance of correctly modelling the interaction between liquid particles and internal element of the combustor to reproduce thermoacoustic oscillations in real configurations has been investigated in a laboratory-scale application aiming to reproduce the physics of real combustion chambers. The results of these analyses have been published in two journal publications (1-3). Furthermore, a joint experimental/numerical study has also been performed in collaboration with EM2C researchers.
The simulations reproduce the experimental observations made in the SICCA-spray test bench developed at the EM2C laboratory (CNRS, CentraleSupelec - Host of secondment in CLEANERFLAMES). The Cerfacs code AVBPS (
http://www.cerfacs.fr/avbp7x(odnośnik otworzy się w nowym oknie)) solving the LES-filtered compressible Navier-Stokes equations was employed for the simulations. Lagrangian formalism is adopted to model the liquid spray. Addressing numerically liquid injection into the combustion chamber, and capturing the specific liquid phase physics, was the first obstacle to overcome: in the developed strategy, primary and secondary atomization phenomena were not explicitly simulated but the semi-empirical model FIM-UR developed at Cerfacs was used to prescribe a droplet diameter distribution at the atomizer orifice that enables to correctly reproduce the spray velocity and diameter distribution downstream the atomization process when the spray is diluted. Two particle-wall interaction treatments were tested and their impact on the spray-flame stabilization and the thermoacoustic combustion instabilities is evaluated: when willing to simulate the thermoacoustic oscillation a “slip wall treatment” was proven to be incapable of reproducing the dynamic response of the liquid fuel impinging on the wall, whereas adopting a “film model” allows to overcome this limit.
A different approach to predict thermoacosutic combustion instabilities relies on the use of reduced order tools (ROM) well-known to provide fast and cheap predictions which are well adapted especially in the engines design phase where numerous modifications have to be tested, e.g. position of the burner, length of the combustion chamber, fuel mass flow rate, etc. These tools address the problem of the thermoacoustic stability of a system from the acoustic point of view where the flames behave as a source/damp of acoustic energy. A proper modelling of the flame as an acoustic element is essential to produce accurate and relevant predictions by use of ROM. This is classically achieved in terms of Flame Transfer Function (FTF) which links the heat release rate perturbation (Q ̇ ) integrated over the flame location with incoming perturbations taken at a reference point. The research activities carried out during the project investigate the use of high-fidelity LES to determine the gain and phase of the FTF for the previously analyzed SICCA-spray combustor. Results of this study have been presented to the international conference “10th European Combustion Meeting”. Modelling was proved to be able to correctly reproduce the flame response but was also found highly sensitive to the injection parameters.
Finally, the capability of a recently develop low-order acoustic network modeling tool STORM (State-space Reduced- Order-Modeling) of handling annular complex geometries, especially to study azimuthal instabilities, were proved through a low-cost linear stability analysis of the MICCA-spray combustor from EM2C. STORM was fond to perform exceedingly well in this context retrieving the stability of all the thermoacoustic modes of the annular combustor in just one calculation within a few minutes.Results of this analysis have been presented to the international conference "European Combustion Meeting". Finally, the proposed modelling is tested on a real SAFRAN aeroengine proving the scalability of the developed methodology to full-scale combustion systems.
