Experiments of relevance to lean-burn gas-turbine combustors were carried out in plane and axisymmetric sudden-expansion flows. They involved the determination of the extinction limits and the nature of combustion as a function of the equivalence rati...
Experiments of relevance to lean-burn gas-turbine combustors were carried out in plane and axisymmetric sudden-expansion flows. They involved the determination of the extinction limits and the nature of combustion as a function of the equivalence ratio of methane-air flames. In addition, they examined the implications for methods of active control of the combustion oscillations observed close to the extinction limits and at equivalence ratios around unity.
The flammability limits and flame characteristics were determined with acoustically-open and closed upstream ends, and with and without exit nozzles. Five distinct regimes of combustion were identified in terms of pressure characteristics and images of CH chemiluminescence. They comprised regions close to the lean and rich extinction limits, one associated with large-amplitude oscillations dominated by longitudinal acoustic or bulk-mode frequencies and observed with near-stoichiometric mixtures, and the other two with nominally stable combustion.
Combustion close to the lean and rich extinction limits led to oscillations with frequencies much lower than those of acoustic and bulk-modes. The emphasis is on these near-limit instabilities since they may be relevant to the combustion of lean mixtures in land-based gas turbines. In the plane duct, they gave rise to low-frequency flapping immediately prior to extinction with the branches of flame behind the steps extinguishing non-simultaneously and lateral and longitudinal when only a single branch remained. Photographs of the CH radical suggested that the oscillations were due to the axial movement of extinction along the shear layer and initiated by the high strain rates close to the step, followed by the upstream propagation of the flame when the strain rate was sufficiently low. Similar oscillations were observed in the round duct but without the flapping. The frequency of the oscillations increased with flow rates and burning velocities and the amplitude increased with heat release and with constriction of the duct exit, with the latter causing the oscillations to couple with the bulk-mode frequency of the combustor cavity. The increase in amplitude led to higher strain rates and the narrowing of the flammability range.
The possibility of controlling the oscillations was examined by imposing pressure oscillations at a frequency different from that of the combustion instability. The imposed oscillations did not affect the near-limit oscillations in the unconstricted ducts and caused a decrease in the flammability range due to the consequent increase in strain rate. In constricted ducts, however, imposed pressure oscillations reduced the amplitudes by up to 70 % because of their ability to suppress the bulk-mode oscillation and, hence, extended the flammability range. Imposed oscillations were also successful with near-stoichiometric oscillations at moderate flow rates but less effective at high flow rates where the signals were modulated by low frequency movement of the flame. Attempts to actively control the near-limit oscillations were unsuccessful due to a cycle-to-cycle variation of around 30 %, but were successfully with the large-amplitude oscillations close to stoichiometry. A new control system was developed to track the signal with on-line actuation to counteract the tendency of the flame to move away from the step.
The effect of imposed and naturally occurring pressure oscillations on NOx emissions is also quantified in view of its importance to lean burn combustors. An increase in the amplitude of pressure oscillations led to an decrease in NOx emissions close to stoichiometry and an increase close to the lean limit, due to the shortening of the residence time at the maximum adiabatic flame temperature and temperature fluctuations above the anticipated local mean temperature, respectively.