The effect of confinement on lean-premixed, swirl stabilized flame response
Open Access
- Author:
- De Rosa, Alexander John
- Graduate Program:
- Mechanical Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- March 04, 2015
- Committee Members:
- Domenic Adam Santavicca, Dissertation Advisor/Co-Advisor
Thomas Litzinger, Committee Member
Richard A Yetter, Committee Member
Sarma V Pisupati, Committee Member
Karen Ann Thole, Committee Member - Keywords:
- combustion dynamics
combustion instability
combustion confinement
instability confinement - Abstract:
- The effect of confinement on the heat release rate response of a fully premixed flame to fluctuations in inlet velocity was investigated experimentally. This investigation was performed using a single-nozzle, swirl-stabilized, gas turbine combustor of industrial design. The global flame response was characterized by a flame transfer function which related fluctuations in the heat release rate from the flame to the artificially imposed fluctuations in the velocity of the mixture entering the combustor. The magnitude of these velocity fluctuations was measured using the two-microphone method while the fluctuations in heat release rate were described by measuring the CH* chemiluminescence emission from the flame. An intensified high speed camera, fitted with a bandpass filter for CH* chemiluminescence was used to described the spatial distribution of the heat release rate within the flame. This investigation was conducted in the linear flame response regime over a wide range of operating conditions; 20-35 m/s inlet velocity, 373-473 K preheat, equivalence ratio 0.55-0.7 and in three combustors of 0.11 m, 0.15 m and 0.19 m diameter. The confinement ratio was 0.5, 0.37 and 0.29 for each combustor when calculated using the diameter of the nozzle relative to the combustor diameter. All measurements were made at atmospheric pressure. The effect of the combustor outer wall temperature on the measured flame structure and response to instability was found to be negligible over the range of wall temperatures assessed. Measurements of the global flame response in terms of the flame transfer function displayed the characteristic series of alternating extrema in the gain and linear phase decay with frequency described in the literature. These trends were also consistent across all combustor diameters. The effect of operating conditions on the measured transfer function in each confinement was found to be similar and was indicative of a similarity in the mechanisms of flame response in each case. The primary effect of confinement on the flame transfer function was found to be an increase in the slope of the phase plot and a reduction in the gain at high frequencies as confinement was increased. Attempts to generalize the entire dataset in terms of the measured flame transfer functions using the Strouhal number were unsuccessful. While the Strouhal number was able to collapse the measured data in a single confinement, the data collected over all three combustor diameters was not collapsed. A study into the parameters that governed the response of the flame in each confinement lead to the introduction of both the confinement ratio (flow expansion ratio), and the ratio of flame length to combustor diameter (flame aspect ratio or flame base angle) into the transfer function normalization. Plotting the collated flame transfer function data against a combination of Strouhal number and confinement ratio was successful in collapsing the 0.15 m and 0.19 m combustor diameter data. Data from the 0.11 m diameter combustor did not collapse. This behavior was attributed to a change in the flow regime as confinement was varied. In the least confined cases the flow was in the free-jet regime and governed by the expansion of the jet from the injector into the combustor. This expansion altered the convective velocity at which velocity perturbations traveled within the system. In the most confined case, it was suggested that the flow was in the wall-jet regime and that the degree of flame-wall interaction, and not the expansion ratio, governed the flow velocity within the system. Flame images were used to further explain the behaviors observed in the global flame response. The shape and structure of the flame was shown to follow a consistent evolution with increasing confinement in that an increase in the degree of flame-wall interaction was found to result in an increase in the axial distribution of the flame's heat release rate. This change was found to occur in both the stable flame shape and the fluctuating flame structure. Furthermore, these changes in location of heat release rate were shown to relate to the changes observed in the transfer function phase and global response of the flame to combustion instability. A new method for describing the different components of flame area typically described in reduced order models was then introduced and used to further analyze the flame images. This technique was able to demonstrate that the position of the flame in the region near the flame anchoring point was independent of confinement. It was also used to recover evidence of convective mechanisms moving along the mean flame position in each confinement case. These convective fluctuations were then shown to be related to a fluctuation in swirl and to be consistent across all combustor sizes. Finally, a new technique for measuring the response of gas turbine flames to instabilities using only the naturally occurring, turbulent fluctuations within the system was described. This technique was able to accurately predict the transfer function gain response of the flame but not the phase. Errors in the measurement technique were associated with the poor resolution of the measurement systems used to acquire the data and several improvements to this method were, therefore, suggested and demonstrated.