COMPUTATIONAL FLUID DYNAMICS (CFD) MODELING OF FLAME TRANSFER FUNCTION IN GAS TURBINE COMBUSTORS

Restricted (Penn State Only)
- Author:
- Wu, Jinming
- Graduate Program:
- Mechanical Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- May 20, 2024
- Committee Members:
- Robert Kunz, Professor in Charge/Director of Graduate Studies
Jacqueline O'Connor, Major Field Member
Daniel Haworth, Major & Minor Field Member
Yuan Xuan, Chair & Dissertation Advisor
Donghyun Rim, Outside Unit & Field Member - Keywords:
- Computational Fluid Dynamics(CFD)
Large Eddy Simulation(LES)
Flame Transfer Function(FTF)
Gas Turbine Combustor - Abstract:
- The flame transfer function (FTF) was used to describe how the premixed flame responds to the inlet acoustic oscillations, which could promote the generation of combustion instabilities in gas turbine combustors. The objective of this research work is to develop a universal FTF calculation framework that is independent of operating conditions and can be applied to higher pressure, different injector, and varying combustor geometries. To achieve this, a robust, standardized FTF calculation framework was developed using Large Eddy Simulations (LES), based on an atmospheric pressure, variable length, swirl-stabilized, lean-premixed gas turbine combustor. The industrial injectors from Solar Turbines, Inc. were employed in this research. A commercial Computational Fluid Dynamics (CFD) software, STAR-CCM+, was implemented for this numerical study. This LES-based FTF calculation tool exhibited consistent predictive capabilities across different operating conditions under 1 atm. Those operating conditions varied by equivalence ratio, mean velocity at the injector exit, injector geometry, and so forth. This tool could capture the overall trend of the FTF phase and gain well when compared with atmospheric measurements. However, it was not able to capture the minimum gain response. In addition, LES performed better than the empirical model in FTF phase prediction, especially at higher forcing frequencies. This empirical model only provided the FTF phase prediction and was trained based on the atmospheric measurements done at Penn State University. This model is a function of flame position and operating conditions (e.g., inlet temperature, injector exit velocity, and so forth). To gain a better understanding of which part of the intricate flame characteristics needed to be captured well for an accurate FTF prediction, numerical experiments were also conducted to investigate the contributions of the outer flame region and main flame characteristics to FTF predictions. It showed that the outer flame region didn’t have significant effects on the FTF phase but had a small influence on FTF gain. Furthermore, the sensitivity of FTF to main flame characteristics was quantified, such as flame length, the center of heat release (COHR), and flame width. To study this, the main flame characteristics were systematically varied by the Turbulent Flame Speed Closure (TFC) model. The results showed FTF phase and gain had low sensitivity to main flame characteristics in the same specific ranges. FTF became very sensitive to flame locations out of those ranges. It suggested that it is not necessary to predict the main characteristics very accurately - as long as those are in the low sensitivity ranges, we can achieve accurate enough FTF predictions. Almost all numerical studies and experiments investigating FTF have focused on combustors operating at atmospheric pressure. However, this differs from the higher pressures typical of real gas turbine combustion systems. To address this gap, the same LES FTF computation tool was utilized at pressures reaching up to 16 atm to evaluate the impact of pressure variations on FTF. Observations from the shape of the unforced flame revealed that increased pressure resulted in a reduction in flame length and width. With a shorter flame under higher pressures, the inlet perturbation takes a shorter time to travel through the flame. This indicates a decreased convective time delay, contributing to a smaller phase delay in the FTF. The FTF results showed that pressure had a notable impact on FTF, given that it significantly alters the main flame characteristics. In addition, to study the effects of injector geometry on FTF, this LES FTF tool was applied to a new industrial injector called T250 which is Solar Turbines, Inc. production model from the Titan 250 (T250) line of combustors. This study demonstrated that the geometry of the injector plays a crucial role in determining the main flame characteristics, which subsequently influence the behavior of the FTF. Moreover, to investigate the effects of combustor geometry on FTF, this LES FTF tool was applied to a new annular combustor with a T70 injector. This rig uses the real operating conditions in industrial applications which will provide useful information for validating our LES FTF calculation framework in real industrial applications. Since this new test rig was using partially premixed gas, the equivalence ratio fluctuation cannot be negligible. So, it is essential to include the equivalence ratio fluctuation in the FTF calculation equation. Based on this new FTF calculation equation, it showed the combustor geometry can have a significant impact on the FTF. This LES-based FTF calculation tool was evaluated under varying operating pressures, injector geometries, and combustor configurations. The findings indicate that while the tool can predict the FTF phase and gain with high accuracy under a range of conditions, the main flame characteristics remains a crucial factor of FTF. The reduced flame length and width at higher pressures and the significant impact of injector and combustor geometry on FTF behavior underscore the necessity for precise modeling in real-world gas turbine applications. This study provides valuable insights for extending FTF predictive capabilities to more diverse and higher-pressure environments, paving the way for more reliable and efficient gas turbine combustor designs.