No abstract

No abstract

No abstract

The flame transfer function in a premixed gas turbine combustor is experimentally determined. The fuel (natural gas) is premixed with air upstream of a choked inlet to the combustor. Therefore, the input to the flame transfer function is the imposed velocity fluctuations of the fuel/air mixture without equivalence ratio fluctuations. The inlet-velocity fluctuations are achieved by a variable-speed siren over the forcing frequency of 75–280 Hz and measured using a hot-wire anemometer at the inlet to the combustor. The output function (heat release) is determined using chemiluminescence measurement from the whole flame. Flame images are recorded to understand how the flame structure plays a role in the global heat release response of flame to the inlet-velocity perturbation. The results show that the gain and phase of the flame transfer function depend on flame structure as well as the frequency and magnitude of inlet-velocity modulation and can be generalized in terms of the relative length scale of flame to convection length scale of inlet-velocity perturbation, which is represented by a Strouhal number. Nonlinear flame response is characterized by a periodic vortex shedding from shear layer, and the nonlinearity occurs at lower magnitude of inlet-velocity fluctuation as the modulation frequency increases. However, for a given modulation frequency, the flame structure does not affect the magnitude of inlet-velocity fluctuation at which the nonlinear flame response starts to appear.

The stability characteristics of a laboratory-scale lean premixed combustor operating on natural gas - hydrogen fuel mixtures have been studied in a variable length combustor facility. The fuel and air were mixed upstream of the choked inlet to the combustor to eliminate equivalence ratio fluctuations and thereby ensure that the dominant instability driving mechanism was flame-vortex interaction. The inlet velocity, inlet temperature, equivalence ratio and percent hydrogen in the fuel were systematically varied, and at each operating condition the combustor pressure fluctuations were measured as a function of the combustor length. The results are presented in the form of two-dimensional stability maps, which are plots of the normalized rms pressure fluctuation versus the equivalence ratio and the combustor length, for a given inlet temperature, inlet velocity, and fuel mixture. In order to understand the effects of operating conditions and fuel composition on the observed stability characteristics, two-dimensional chemiluminescence images of the flame structure were recorded at all operating conditions and for all fuel mixtures under stable conditions. Changes in the stable flame structure, as characterized by the location of the flame’s “center of heat release”, were found to be consistent with the observed instability characteristics. The location of the flame’s “center of heat release” was found to lie along a single path for all operating conditions and fuel mixtures. It was also observed that there were regions of stable and unstable combustion as one moved along this path. Furthermore it was found that flames having the same “center of heat release” location, but different operating conditions and fuel composition, have very nearly the same flame shape. These results will be useful for developing phenomenological models for predicting unstable combustion.

Flame transfer function measurements of turbulent premixed flames were made in a model lean premixed, swirl-stabilized, gas turbine combustor. OH*, CH*, and CO2* chemiluminescence emissions were measured to determine heat release oscillation from a whole flame, and the two-microphone technique was used to measure inlet velocity fluctuation. 2-D CH* chemiluminescence imaging was used to characterize the flame shape: the flame length (LCH* max) and flame angle (α). Using H2-natural gas composite fuels, XH2 = 0.00 ∼ 0.60, very short flame was obtained and hydrogen enrichment of natural gas had a significant impact on the flame structure and flame attachment points. For a pure natural gas flame, the flames exhibit a “V” structure, whereas H2-enriched natural gas flames have an “M” structure. Results show that the gain of “M” flames is much smaller than that of “V” flames. Similar to results of analytic and experimental investigations on the flame transfer function of laminar premixed flames, it shows that the dynamics of a turbulent premixed flame is governed by three relevant parameters: the Strouhal number (St = LCH* max / Lconv), the flame length (LCH* max), and the flame angle (α). Two flames with the same flame shape exhibit very similar forced responses, regardless of their inlet flow conditions. This is significant because the forced flame response of a highly turbulent, practical gas turbine combustor can be quantitatively generalized using the non-dimensional parameters which collapse all relevant input conditions into the flame shape and the Strouhal number.

Flame transfer function measurements of turbulent premixed flames are made in a model lean-premixed, swirl-stabilized, gas turbine combustor. OH∗, CH∗, and CO2∗ chemiluminescence emissions are measured to determine heat release oscillation from a whole flame, and the two-microphone technique is used to measure inlet velocity fluctuation. 2D CH∗ chemiluminescence imaging is used to characterize the flame shape: the flame length (LCH∗ max) and flame angle (α). Using H2-natural gas composite fuels, XH2=0.00–0.60, a very short flame is obtained and hydrogen enrichment of natural gas is found to have a significant impact on the flame structure and flame attachment points. For a pure natural gas flame, the flames exhibit a “V” structure, whereas H2-enriched natural gas flames have an “M” structure. Results show that the gain of M flames is much smaller than that of V flames. Similar to results of analytic and experimental investigations on the flame transfer function of laminar premixed flames, it shows that the dynamics of a turbulent premixed flame is governed by three relevant parameters: the Strouhal number (St=LCH∗ max/Lconv), the flame length (LCH∗ max), and the flame angle (α). Two flames with the same flame shape exhibit very similar forced responses, regardless of their inlet flow conditions. This is significant because the forced flame response of a highly turbulent, practical gas turbine combustor can be quantitatively generalized using the nondimensional parameters, which collapse all relevant input conditions into the flame shape and the Strouhal number.

Flame response to imposed velocity fluctuations is experimentally measured in a single-nozzle turbulent swirling fully-premixed combustor. The flame transfer function is used to quantify the flame's response to imposed velocity fluctuations. Both the gain and phase of the flame transfer function are qualitatively similar for all operating conditions tested. The flame transfer function gain exhibits alternating regions of decreasing gain with increasing forcing frequency followed by regions of increasing gain with increasing forcing frequency. This alternating behavior gives rise to gain extrema. The flame transfer function phase magnitude increases quasi-linearly with increasing forcing frequency. Deviations from the linear behavior occur in the form of inflection points. Within the field, the current understanding is that the flame transfer function gain extrema are caused by the constructive/destructive interference of swirl number fluctuations and vortex shedding. Phase-synchronized images of forced flames are acquired to investigate the presence/importance of swirl number fluctuations, which manifest as fluctuations in the mean flame position and vortex shedding in this combustor. An analysis of phase-synchronized flame images reveals that mean flame position fluctuations are present at forcing frequencies corresponding to flame transfer function gain minima but not at forcing frequencies corresponding to flame transfer function gain maxima. This observation contradicts the understanding that flame transfer function gain maxima are caused by the constructive interference of mean flame position fluctuations and vortex shedding, since mean flame position fluctuations are shown not to exist at flame transfer function gain maxima. Further analysis of phase-synchronized flame images shows that the variation of the mean flame position fluctuation magnitude with forcing frequency follows an inverse trend to the variation of flame transfer function gain with forcing frequency, i.e., when the mean flame position fluctuation magnitude increases, the flame transfer function gain decreases and vice versa. Based on these observations it is concluded that mean flame position fluctuations are a subtractive effect. The physical mechanism through which mean flame position fluctuations decrease flame response is through the interaction of the flame with the Kelvin–Helmholtz instability of the mixing layer in the combustor. When mean flame position fluctuations are large the flame moves closer to the mixing layer and damps the Kelvin–Helmholtz instability due to the increased kinematic viscosity, fluid dilatation, and baroclinic production of vorticity with the opposite sign associated with the high temperature reaction zone.