Identification of Combustion Noise and Flame Dynamics of Confined Turbulent Flames
Acoustics
Technical Thermodynamics
Final Report Abstract
Research work was carried out by the group of Prof. Wolfgang Polifke at TUM in close collaboration with the group of Prof. Schuller at Laboratoire EM2C, Université Paris-Saclay, France. Combustion noise is an undesirable but unavoidable by-product of every applied turbulent combustion device that may trigger thermoacoustic instabilities, which, in turn, limit the operational conditions of a combustion device. In turbulent confined combustion systems, as they are found in most applied devices, the thermoacoustic stability and the noise levels reached are determined by an interplay of three main components: the combustion noise source, the flame dynamic response to acoustic perturbations and the cavity acoustics. The project dealt with the thermoacoustic modeling of confined combustion systems. Large eddy simulation was combined with advanced system identification techniques to infer models for the flame dynamic response and the combustion noise source. By incorporating the identified models into a linear acoustic network model description of the combustor acoustics, predictions of linear stability margins, combustor dynamics and sound pressure distributions were achieved. An extensive comparison was conducted between the results obtained and experimental data provided from Laboratoire EM2C, Université Paris-Saclay. Via a reactive compressible large eddy simulation experimentally measured sound pressure spectra of a confined lab-scale combustor were precisely reproduced. Thereby, qualitative and quantitative agreement was found not only for thermoacoustically stable, but also for intermittent and fully unstable working conditions. Advanced system identification techniques were investigated and evaluated, which allowed to extend the established large eddy simulation / system identification approach: not only a model for the flame dynamic response was identified, but also an estimation of the combustion noise source was obtained from the same time series data. Surrogate data studies proved that compared with the established finite impulse response identification, a Box-Jenkins model yielded unbiased estimates for the flame dynamic response in case of a closed-loop identification and additionally provided a noise model estimate from the same time series data. Combining broadband time series data from the validated large eddy simulation and the investigated Box-Jenkins identification, models were identified for the flame dynamic response and the combustion noise source. Incorporation of these models into a linear acoustic network model allowed predictions of the combustor dynamics and its sound pressure distribution with reasonable computational effort. The employ of the reduced order model yielded excellent agreement with experimental values of the combustor scattering matrix and the sound pressure spectra. It allowed an intuitive physics-based interpretation of the spectra in terms of cavity resonances and acoustic-flame coupling. It was proven that the incorporation of the flame response into the network model is mandatory for a correct sound pressure prediction. In summary, a coherent framework was presented for the investigation of combustor dynamics and combustor noise: broadband data were generated by a compressible large eddy simulation, which was post-processed via advanced system identification techniques to infer simultaneously models for the flame dynamic response and the generation of combustion noise. In the conducted project these models were combined with a linear acoustic network model to obtain a reduced order model that allowed predictions of combustor dynamics and combustor noise across a large parameter space with reasonable computational costs.
Publications
- The Contribution of Intrinsic Thermoacoustic Feedback to Combustion Noise and Resonances of a Confined Turbulent Premixed Flame. Combustion and Flame, 182:269–278, 2017
C. F. Silva, M. Merk, T. Komarek, and W. Polifke
(See online at https://doi.org/10.1016/j.combustflame.2017.04.015) - Direct Assessment of the Acoustic Scattering Matrix of a Turbulent Swirl Combustor by Combining System Identification, Large Eddy Simulation and Analytical Approaches. Journal of Eng. for Gas Turbines and Power, 2018
M. Merk, R. Gaudron, C. Silva, M. Gatti, C. Mirat, T. Schuller, and W. Polifke
(See online at https://doi.org/10.1115/1.4040731) - Identification of flame transfer functions in the presence of intrinsic thermoacoustic feedback and noise. Combustion Theory and Modelling, 22(3):613–634, 2018
S. Jaensch, M. Merk, T. Emmert, and W. Polifke
(See online at https://doi.org/10.1080/13647830.2018.1443517) - Measurement and Simulation of Combustion Noise and Dynamics of a Confined Swirl Flame. AIAA Journal, 56(5):1930–1942, 2018
M. Merk, R. Gaudron, M. Gatti, C. Mirat, T. Schuller, and W. Polifke
(See online at https://doi.org/10.2514/1.J056502) - Prediction of Combustion Noise of an Enclosed Flame by Simultaneous Identification of Noise Source and Flame Dynamics. Proceedings of the Combustion Institute, 37, 2018
M. Merk, R. Gaudron, C. Silva, M. Gatti, C. Mirat, T. Schuller, and W. Polifke
(See online at https://doi.org/10.1016/j.proci.2018.05.124) - Simultaneous identification of transfer functions and combustion noise of a turbulent flame. Journal of Sound and Vibration, 422:432–452, 2018
M. Merk, S. Jaensch, C. Silva, and W. Polifke
(See online at https://doi.org/10.1016/j.jsv.2018.02.040)
