Identification of premixed Flame Dynamics by transient CFD
by T. Komarek, L. Tay Wo Chong Hilares and Wolfgang Polifke
Introduction
Thermo-acoustic instabilities are a long known phenomenon in technical combustion systems. The introduction of lean premix combustion technology became a significant problem for developers of gas turbines as its implementation generates instabilities. The Instabilities arise out of an unfavourable coupling of acoustics and combustion, developing a self enhancing closed cycle. This can lead to very strong oscillations, reduced efficiency, increased pollution and might even result in damaged parts.
In order to develop reliable prediction tools for instabilities, the acoustic properties of all elements of a new system have to be known. As the complex interaction between acoustics, flow and combustion is not yet fully understand, the dynamics of the combustion in a gas turbine can not be sufficiently modelled. A new method uses system identification theory to identify the acoustic characteristics of the flame from the data of a transient CFD simulation.
Project
The aim of this project is the validation of this method and the investigation of 3D flame dynamics in an annular gas turbine.
A swirl stabilized premixed burner with an axial swirl generator is used as validation case. The burner is investigated experimentally as well as numerically.
The acoustic properties of a flame are commonly described with the flame transfer function. It describes the reaction of the flame to an excitation by acoustic velocity fluctuations in axial direction.
Experiment
During first measurements, the flow field in the combustion chamber and the heat release of the flame were obtained. These results are used for the validation of the steady state simulation.
CFD Simulation
Results of the steady state CFD simulation are displayed in Figure 3. Transient simulations are used to obtain the data for the identification of the flame transfer function.
3D Effects
In an annular combustion chamber higher order modes develop and induce azimuthal velocity fluctuations. The flame is commonly modelled with a flame transfer function, which does not take into account any response to azimuthal excitations. In this project it is investigated if this simplification is justified or if a different model should be used.
One possible way to modify the flame transfer function (F) would be to define it as a superposition of independent responses to different velocity fluctuations.
The investigation is conducted with the TD1 burner that was also developed at the chair. The complete burner with a segment of the annular combustion chamber and of the plenum is simulated (Figure 4). The system is excited with axial and azimuthal velocity fluctuations. In the post processing, dependencies of the flame dynamics are identified.
Outlook
The next challenges in this project are:
- Measurements of the flame transfer function of the BRS burner.
- Reconstruction of the flame transfer function of the BRS burner from the data of transient CFD simulations.
- Numerical investigation of the flame response with azimuthal excitation in an annular combustion.