Flame transfer function measurements and mechanisms in a single-nozzle combustor
Open Access
- Author:
- Bunce, Nicholas Allen
- Graduate Program:
- Mechanical Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- June 14, 2013
- Committee Members:
- Domenic Adam Santavicca, Dissertation Advisor/Co-Advisor
Daniel Connell Haworth, Committee Member
Savas Yavuzkurt, Committee Member
Michael Matthew Micci, Committee Member - Keywords:
- combustion
gas turbine
flame response
swirling - Abstract:
- The response of a fully-premixed flame to velocity fluctuations was experimentally measured in a single-nozzle, swirl-stabilized, model gas turbine combustor. Flame response was quantified in terms of the flame flame transfer function relating the input velocity fluctuations to the output heat release rate fluctuations. The velocity fluctuation was measured using the two-microphone method and the heat release rate fluctuation was measured using $CH^*$ chemiluminescence emission over the forcing frequency range of $100-500\text{ Hz}$ with a fixed velocity fluctuation amplitude, $\sfrac{u^{\prime}_{rms}}{\overline{u}}$, of 5\%. Measurements were conducted over a broad range of operating conditions encompassing varied combustor pressure, $0.1-0.4\text{ MPa}$, inlet temperature, $373-573\text{ K}$, average velocity, $15-35\text{ }\sfrac{m}{s}$, and equivalence ratio, $0.45-0.75$. A total of 47 flame transfer function measurements were acquired over this range of operating conditions. Time-averaged $CH^*$ chemiluminescence flame images were acquired at all operating conditions. At select operating conditions, the flame structure during forcing was characterized through high-speed $CH^*$ chemiluminescence flame imaging. Flame transfer function gain at all operating conditions exhibited similar characteristics indicating that the same velocity fluctuation mechanisms may be present at all operating conditions. At low frequencies, flame transfer function gain decreased with increasing forcing frequency. After reaching a minimum, flame transfer function gain then increased with increasing forcing frequency. Once a maximum was reached the behavior repeated. Flame transfer function phase increased quasi-linearly with increasing forcing frequency. Deviation from the linear trend occurred in the form of inflection points at forcing frequencies corresponding to flame transfer function gain minima. The effect of each operating condition parameter on the flame transfer function was investigated independently. Velocity fluctuation mechanisms were investigated from a global perspective by comparing the collapse of flame transfer function gain with different frequency scaling parameters. Four frequency scaling parameters were compared: Strouhal number based on flame length $\left(St_{L_{fl}}\right)$, Strouhal number based on nozzle diameter $\left(St_{D_{nozzle}}\right)$, phase between axial and azimuthal velocity fluctuations at the flame anchoring location $\left(\theta_{v-u}\right)$, and phase between swirl number and axial velocity fluctuations at the flame anchoring location $\left(\theta_{S-u}\right)$. It was found that $\left(\theta_{v-u}\right)$ collapsed the flame transfer function gain best. Since this parameter is directly related to the swirl number fluctuation magnitude it indicates that swirl number fluctuations are an important velocity fluctuation mechanism. It was also found that the maximum flame transfer function gain decreased with increasing $\sfrac{S_L}{\overline{u}}$ which is related to the response time of the flame. Velocity fluctuation mechanisms were then investigated on a local scale through analysis of phase-synchronized flame images. Root mean square fluctuation images showed that heat release fluctuations are equally distributed about the mean flame position at flame transfer function gain minima. Conversely, at flame transfer function gain maxima the largest heat release fluctuation occurred in the downstream region of the flame. A windowing analysis was applied to the phase-synchronized flame images to investigate the interference of axial velocity and swirl number fluctuations. It was found that interference between these two mechanisms was only present at flame transfer function gain minima, and then only for certain window divisions showing that interference between between the two mechanisms is not the cause of the flame transfer function gain extrema. Swirl number fluctuations were then examined through their direct effect on the flame, movement of the flame base position. Flame base movement followed an inverse trend to flame transfer function gain, i.e. when flame transfer function gain increased flame base movement decreased and vice versa. This trend was shown for all but the shortest flames tested. This indicates that flame base movement acts to decrease global flame response and that the degree of flame wall interaction modifies flame response. Through examination of the vorticity equation it was shown how the flame could decrease the vorticity of the flow by gas expansion, baroclinic production of vorticity of opposite side, and increased viscous diffusion. Therefore it is proposed that when the swirl number fluctuation is largest the flame base movement is largest and the position of the flame relative to the shear layer changes causing decreased vorticity and in turn decreased flame transfer function gain. When the swirl number fluctuation is smallest the flame base does not move and the vorticity of the shear layer is not damped before interacting with the flame leading to high flame transfer function gain.