Modeling and Identification of Technically Premix Flame Dynamics

• Applicants: Professor Dr. Michael Pfitzner; Professor Wolfgang Polifke, Ph.D.
• Subject Area: Energy Process Engineering ; Fluid Mechanics ; Technical Thermodynamics
• Term: from 2017 to 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 393001638

In order to study self-excited combustion instabilities, it is essential to characterize the response of the flame of interest to relevant flow perturbations. In the proposed study, the dynamic response of technically premixed flames to perturbations of upstream velocity and equivalence ratio will be modelled and quantified simultaneously. This will be achieved by combining validated Large Eddy Simulation (LES) of turbulent, reacting flow with advanced System Identification (SI) in order to generate reduced order models of flame dynamics that properly respect the Multiple-Input / Single-Output (MISO) structure of technically premixed flames. Two main challenges will be addressed: 1) Advanced SI methods, employing Box-Jenkins models with MISO structure, will be used to identify reduced order response functions from LES time series data. Unlike models based on traditional identification schemes, the resulting flame models will be independent of the acoustic properties of the fuel injection system. In addition, combustion noise resulting from resolved turbulent fluctuations will not be disregarded in the advanced SI framework, as was done previously, but instead will be modelled as colored noise. This enhances the quality of the overall identification and allows to quantify uncertainties that result from finite time series length. 2) Reduced order models determined with System Identification (SI) can only be as accurate as the time series data that they are deduced from. Furthermore, MISO identification requires time series of considerable length. Thus a turbulent combustion model for LES of technically premixed flames that is both accurate and efficient must be employed, or developed and validated with respect to mean and fluctuating quantities. Furthermore, the capability of LES combustion models to predict flame dynamics will be validated. The model must be reliable even for coarse meshes and at elevated pressure, where sub-grid flame wrinkling can be considerable. The MISO models determined with validated LES/SI can be combined with acoustic models to study, for example, the impact of location and acoustic impedance of fuel injectors on the thermoacoustic stability of combustors in a flexible and efficient manner.

• DFG Programme: Research Grants