Direct and Indirect Sources in Combustion for Evaluation and Reduction of Noise – DISCERN

The experimental test bench is representative of practical situations associated with new generation engines. Indeed, the flow is highly swirled and the combustion is premixed. Moreover, the combustion chamber is short and equipped with a nozzle that is adapted to combustion noise study. Boundary conditions are well controlled. Numerous sensors and diagnostics are available and can be synchronized, enabling a precise study of combustion and burnt gas dynamics.
In this context, the most adapted prediction method is large eddy simulation (LES). The literature shows that near-wall pressure fluctuations can be properly calculated by LES solvers in regimes dominated by combustion instabilities (strong acoustic coupling). Predicting correctly the combustion noise levels in these systems in the absence of instability remains a challenge and the simulations proposed here allow advanced comparisons between experiments and calculations.
Reduced-order models (ROM) are developed based on experimental and numerical results. After validation, simulations are used to establish the data and parameters that are necessary to properly calibrate ROMS.

We developed an experimental database for the validation of combustion noise prediction methods. This database contains data that can be directly used for comparison with numerical or modelling results. We have conducted extremely precise large eddy simulations of the experimental system and demonstrated the predictive capabilities of LES in this field. We identified direct and indirect noise sources and improved our knowledge of these phenomena. We have proposed a reduced order
methodology for the estimation of combustion noise. These three successes made it possible for the partners to integrate RECORD, a major European program.

The DISCERN project presents various and numerous innovating aspects. In particular, we were able to precisely determine the repartition of acoustic sources in the combustion chamber and to optimize on the acoustical point of view the associated nozzle geometry.

Comparing experiments, simulations and modelling allowed us to write several joint papers, as can be seen in the list of publications and communications of the present report.

The present proposal aims at characterizing the direct and indirect combustion noise sources involved in sound emission in industrial processes, with a particular focus on aero- and power gas turbines. This is done using in parallel experimental, numerical and theoretical approaches. Experimentally, combustion noise is generated and measured on a dedicated combustor including the main features of practical situations of interest: a lean partially premixed turbulent flame stabilized by swirl in a pressurized chamber. In parallel, calculations are carried out on this configuration with the high fidelity large eddy simulation (HFLES) solver AVBP. The first objective is to recover the measured acoustic pressure levels and spectra, validating both strategies in the same time. Complementary diagnostics will then be used on trustful HFLES and experimental databases to extract and clearly discern direct and indirect contributions to the noise emitted in the chamber. Meanwhile, low-order simulations of the downstream part of the chamber, including the nozzle, will be performed to investigate the influence of the nozzle geometry on the indirect combustion noise generation and transmission. The project involves three partners with a wide range of complementary skills on the different approaches. The final objective is to provide a good understanding and reliable models for the prediction and reduction of combustion noise in practical situations. This proposal has obtained the INCA and IROQUA labels. We are also presently applying for the “pôles de compétitivité” ASTECH (Ile-de-France) and AESE (Sud-Ouest) labels. Recommendation letters can be provided if asked.