A new approach was developed to determine the operational stability margin of a laboratory scale combustor. Applying modern and robust techniques and tools from Dynamical System Theory, the approach was based on three basic steps. In the first step, a gray-box thermoacoustical model for the combustor was derived. The second step consisted in applying System Identification techniques to experimental data in order to validate the model and estimate its parameters. The application of these techniques to experimental data under different operating conditions allowed us to determine the functional dependence of the model parameters upon changes in an experimental control parameter. Finally, the third step consisted in using that functional dependence to predict the response of the system at different operating conditions and, ultimately, estimate its operational stability margin. The results indicated that a low-order stochastic non-linear model, including two excited modes, has been identified and the combustor operational stability margin could be estimated by applying a continuation method.

Low swirl burners (LSBs) have gained popularity in heating and gas power generation industries, in part due to their proven capacity for reducing the production of NOx, which in addition to reacting to form smog and acid rain, plays a central role in the formation of the tropospheric ozone layer. With lean operating conditions, LSBs are susceptible to combustion instability, which can result in flame extinction or equipment failure. Extensive work has been performed to understand the nature of LSB combustion, but scaling trends between laboratory- and industrial-sized burners have not been established. Using hydrogen addition as the primary method of flame stabilization, the current work presents results for a 2.54 cm LSB to investigate potential effects of burner outlet diameter on the nature of flame stability, with focus on flashback and lean blowout conditions. In the lean regime, the onset of instability and flame extinction have been shown to occur at similar equivalence ratios for both the 2.54 cm and a 3.81 cm LSB and depend on the resolution of equivalence ratios incremented. Investigations into flame structures are also performed. Discussion begins with a derivation for properties in a multicomponent gas mixture used to determine the Reynolds number (Re) to develop a condition for turbulent intensity similarity in differently-sized LSBs. Based on this requirement, operating conditions are chosen such that the global Reynolds number for the 2.54 cm LSB is within 2% of the Re for the 3.81 cm burner. With similarity obtained, flame structure investigations focus on flame front curvature and flame surface density (FSD). As flame structure results of the current 2.54 cm LSB work are compared to results for the 3.81 cm LSB, no apparent relationship is shown to exist between burner diameter and the distribution of flame surface density. However, burner diameter is shown to have a definite effect on the flame front curvature. In corresponding flow conditions, a decrease in burner diameter results a broader distribution of curvature and an increased average curvature, signifying that compared to the larger 3.81 cm LSB, the flame front of the smaller burner contains tighter, smaller scale wrinkling.

<div>The continued advancement of gas turbine combustion technology for power generation and propulsion applications requires novel techniques to increase the overall engine cycle efficiency and improved methods for mitigating combustion instabilities. To help address these problems, high-speed optical diagnostics were applied to two different experiments that replicate relevant physics in gas turbine combustors. The focus of the measurements was to elucidate the effect of various operating parameters on combustion dynamics occurring over a wide range of spatio-temporal flow and chemical scales. The first experiment, VIPER-M, enabled the investigation of coupling mechanisms for transverse instabilities in a multi-element, premixed combustor that maintains key similarities with gas turbine combustors for land based power generation. The second experiment, COMRAD, facilitated the study of the effect of fuel heating on the combustion performance and dynamics in a liquid-fueled, piloted swirl flame typical of aviation engine combustors. </div><div> </div><div><br></div><div>Two different injector lengths were tested in the VIPER-M experiment, and high-speed CH* chemiluminescence imaging and an array of high-frequency pressure transducers were used to characterize the overall combustor dynamics. For all conditions tested, the longer injector length configuration exhibited high-amplitude instabilities, with pressure fluctuations greater than 100% of the mean chamber pressure. This was due to the excitation of the fundamental transverse mode, with a frequency around 1800 Hz, as well as multiple harmonics. Shortening the injector length significantly lowered the instability amplitudes at all conditions and excited an additional mode near 1550 Hz for lower equivalence ratio cases. The delineating feature controlling the growth of the instabilities in the two injector configurations was shown to be the coupling between the transverse modes in the chamber and axial pressure fluctuations in the injectors.</div><div> </div><div><br></div><div>Heated fuels were introduced into the COMRAD experiment, and simultaneous 10 kHz stereoscopic particle image velocimetry and OH* chemiluminescence imaging were performed over a range of equivalence ratios and combustor pressures to study the influence of fuel temperature on the flow and flame structure. The main flame was found to move upstream as the fuel was heated, while no changes in the pilot flame location were observed in the field of view at the exit of the injector. The upstream shift of the main flame corresponded to a local increase in the axial velocity, which caused the shear layer between the pilot/main flames and the central recirculation zone to move downstream. Direct comparison of the mean velocity fields relative to the mean flame location showed that heating the fuel caused the velocity normal to the flame front to increase, which is indicative of an increase in flame speed. The changes to the fuel injection and chemical kinetics help explain the local changes to the flow and flame structure, which contribute to an overall increase in combustion efficiency as well as NO_x emissions.</div><div> </div><div><br></div><div>Lastly, the effect of fuel injection temperature on the presence of an 800 Hz combustion instability in the COMRAD experiment was investigated. High-frequency pressure and high-speed chemiluminescence measurements revealed a decrease in the instability amplitude as the fuel was heated. The coupling between the fuel flow and the unsteady heat release was studied using independent 10 kHz stereoscopic particle image velocimetry and 10 kHz Mie scattering measurements. The variations in the fuel flow entering the combustor over the acoustic cycle decreased as the instability amplitude weakened. 100 kHz burst-mode, two-component particle image velocimetry was then applied to the unstable condition with ambient temperature fuel. This measurement was capable of resolving both the large-scale changes to the structure of the inner recirculation zone occurring at 800 Hz as well as the time-evolution of small-scale vortex structures. The vortices were shown to correspond to a characteristic frequency in the range of 4-5.5 kHz, and the strength of the vortex structures fluctuated with the global 800 Hz combustion dynamics. These results highlight the importance of performing measurements capable of resolving the wide range of scales present in the flow-fields typical of gas turbine combustors to improve current understanding of flame-flow coupling mechanisms.</div>

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第19685号 / 工博第4140号 / 新制||工||1639(附属図書館) / 32721 / 京都大学大学院工学研究科機械理工学専攻 / (主査)教授 小森 悟, 教授 中部 主敬, 教授 稲室 隆二 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.