Flame Dynamics and Stability Analysis
Motivation
Two recent aspirations in the development of next generation gas turbines are the reduction of pollutant emissions and the enhancement of their operational flexibility. Both are essentially driven by the increasing relevance of renewable energy sources. The erratic availability of wind and sun light generates a need for compensation. Thus, power plants - and gas turbines in particular - need to be started up more frequently, their operating range has to be extended and a quicker change of operating points should be feasible. At the same time emissions, such as NOx, have to be as low as possible.
At the Professur für Thermofluiddynamik research focuses on gas turbine combustors. It has been shown that above-mentioned requirements are best met by turbulent premixed combustion technologies. Those allow considerable control of the combustion process, however, they are prone to so called thermoacoustic instabilities. They arise when the transient acoustic field within the combustion chamber interacts in a reinforcing way with the unsteady heat release of the flame. This could lead to unstable operation and even to failure or damage of the gas turbine.
In order to build efficient and flexible combustors, it is required to enlarge stability margins and ensure a save and stable operation close to unstable operating points. Thus we need a fundamental understanding of the dynamics of thermoacoustic instabilities which would enable us to develop more precise and faster methods to predict and prevent the onset of high amplitude oscillations.
Objectives and Strategy
Combustor acoustics can be represented by so called network models. All relevant components - like ducts, orifices or flames - can be characterized by their input-output behavior. This can be done by identifying transfer functions. Subsequently these modules can be connected in parallel or in series to networks. This project primarily focuses on modeling the flame part and its influence on the system stability, especially the so called flame intrinsic feedback.
In the case of turbulent combustion many processes happen in parallel and affect each other. For that reason in a first step only laminar flames are investigated. Here, cause and effect can be separated in a more clear way. This would pave the way for the development of a parametric reduced order model (ROM) which give decent predictions of the flame transfer function.
Methods
For validation and model development a comprehensive data base is necessary. Therefore 2D CFD simulations of laminar premixed flames are carried out using different solvers (compressible and incompressible) which are cross-validated against each other as well as against experimental investigations. The flames are excited by applying a broad band or a step signal to the inlet velocity profile (see figure 1). Best-practice guidelines are desired.
The reduced order modeling is done using the G-equation formalism which treats the flame as a gas dynamic discontinuity. A lot of work has already been published in this specific field of research, however, it still lacks in a physically motivated velocity model. Our approach is to decompose the velocity field into a potential and a solenoidal part and to model each one separately. This reduces complexity and allows to derive analytic models.