Modeling and Identification of Technically Premix Flame Dynamics
Fluid Mechanics
Technical Thermodynamics
Final Report Abstract
Despite the restrictions due to the Covid-19 pandemic and delays recruiting the PhD candidates meaningful results have been created helping to improve the understanding of several aspects of modeling and identification of technically premixed flame dynamics. Two different types of combustion models (ATF and FSD) were extensively validated and compared with experimental data. The ATF model proved to be a valuable combustion model due to it’s simplicity and good results "out-of-the-box". The model is less sensitive to the turbulence sub-grid model and amount of resolved turbulent kinetic energy. Due to its increased diffusivity, the model is numerically very robust. Besides, the explicit calculation of the chemical source terms consider cooling effects on the chamber walls. The FSD model could be successfully extended to technically premixed flames. Due to it’s modeling approach of the combustion process, there is a stronger dependency on the chosen model parameters, which must be carefully set depending on, e.g. the turbulence model and grid resolution. Due to its modular structure for different physical effects, it can be adapted independently. Unfortunately, the available quenching models were not sufficient to completely account for wall quenching effects. Surprisingly, the FSD model showed very good results for the technically premixed test case without case-dependent model parameter fitting. However, this test case is without swirl inlets and flame-wall interaction. The simulations of dynamic flames with flame-wall interaction showed that these two aspects need to be carefully addressed for the FSD model. The new oblique combustion model is able to predict the filtered reaction source term of a single progress variable for LES-typical filter sizes. Promising results could be achieved with the new model but no a posteriori results could be achieved due to time limitations. The comparison of steady-state and dynamic results revealed that accurate steady-state results do not guarantee satisfactory prediction of the flame dynamics and vice versa. Interestingly, in some cases models predicted flames with a wrong flame shape or a significantly shortened flame length but provided better flame dynamic predictions, than models predicting flames with more accurate steady-state mean fields (especially for the FSD model). This fact could be ascribed to a wrong prediction of cooled wall quenching effects for the FSD model. However, it underlines the importance of evaluating steady-state as well as dynamic results. System Identification methods were improved and extended for technically premixed systems. A significant reduction of return time and computation resources at a given accuracy threshold compared to classical broad-band based LES/SI was achieved through a novel data fusion and system level parallelization scheme. Two identification approaches, i.e. double SISO and direct MISO, were developed and successfully applied in LES/SI of the PRECCINSTA burner. The double SISO identification approach reduces the requirements on experimental equipment and setup. Both methods are able to quantify the uncertainty of the model with similar results. Interestingly, the inclusion of a noise model did not further increase the identification quality.
Publications
-
A New Analytic PDF for Simulations of Premixed Turbulent Combustion. Flow, Turbulence and Combustion 106, 4 (2021), 1213–1239
Pfitzner, M.
-
A top level parallelization and data fusion approach for identification of flame transfer functions with increased reliability, accuracy and efficiency. In 27th International Congress on Sound and Vibration 2021 (ICSV27) (Online, July 2021), International Institute of Acoustics and Vibration (IIAV)
Kuhlmann, J., Guo, S., and Polifke, W.
-
An analytic probability density function for partially premixed flames with detailed chemistry. Physics of Fluids 33, 3 (2021), 035117
Pfitzner, M., and Breda, P.
-
Investigation of pressure and the Lewis number effects in the context of algebraic flame surface density closure for LES of premixed turbulent combustion. Theoretical and Computational Fluid Dynamics 35, 1 (2021), 17–37
Allauddin, U., Lomada, S. R. R., and Pfitzner, M.
-
A multidimensional combustion model for oblique, wrinkled premixed flames. Combustion and Flame 241 (2022), 112121
Pfitzner, M., Shin, J., and Klein, M.
-
Assessing accuracy, reliability and efficiency of combustion models for prediction of flame dynamics with large eddy simulation. Physics of Fluids 34, 9 (2022), 095117
Kuhlmann, J., Lampmann, A., Pfitzner, M., and Polifke, W.
-
LES-based prediction of technically premixed flame dynamics and comparison with perfectly premixed mode. Physics of Fluids 34, 8 (2022), 085125
Kuhlmann, J., Marragou, S., Boxx, I., Schuller, T., and Polifke, W.
