No abstract

Lean premixed prevaporised (LPP) combustion can reduce NOx emissions from gas turbines, but often leads to combustion instability. A flame transfer function describes the change in the rate of heat release in response to perturbations in the inlet flow as a function of frequency. It is a quantitative assessment of the susceptibility of combustion to disturbances. The resulting fluctuations will in turn generate more acoustic waves and in some situations self-sustained oscillations can result. Flame transfer functions for LPP combustion are poorly understood at present but are crucial for predicting combustion oscillations. This paper describes an experiment designed to measure the flame transfer function of a simple combustor incorporating realistic components. Tests were conducted initially on this combustor at atmospheric pressure (1.2 bar and 550 K) to make an early demonstration of the combustion system. The test rig consisted of a plenum chamber with an inline siren, followed by a single LPP premixer/duct and a combustion chamber with a silencer to prevent natural instabilities. The siren was used to induce variable frequency pressure/acoustic signals into the air approaching the combustor. Both unsteady pressure and heat release measurements were undertaken. There was good coherence between the pressure and heat release signals. At each test frequency, two unsteady pressure measurements in the plenum were used to calculate the acoustic waves in this chamber and hence estimate the mass-flow perturbation at the fuel injection point inside the LPP duct. The flame transfer function relating the heat release perturbation to this mass flow was found as a function of frequency. The same combustor hardware and associated instrumentation were then used for the high pressure (15 bar and 800 K) tests. Flame transfer function measurements were taken at three combustion conditions that simulated the staging point conditions (Idle, Approach and Take-off) of a large turbofan gas turbine. There was good coherence between pressure and heat release signals at Idle, indicating a close relationship between acoustic and heat release processes. Problems were encountered at high frequencies for the Approach and Take-off conditions, but the flame transfer function for the Idle case had very good qualitative agreement with the atmospheric-pressure tests. The flame transfer functions calculated here could be used directly for predicting combustion oscillations in gas turbine using the same LPP duct at the same operating conditions. More importantly they can guide work to produce a general analytical model.

Auto-ignition delay time measurements have been attempted for a variety of gaseous fuels on a flow rig at gas turbine relevant operating conditions. The residence time of the flow rig test section was approximately 175 ms. A chemical kinetic model has been used in Senkin, one of the applications within the Chemkin package, to predict the auto-ignition delay time measured in the experiment. The model assumes that chemistry is the limiting factor in the prediction and makes no account of the fluid dynamic properties of the experiment. Auto-ignition delay time events were successfully recorded for ethylene at approximately 16 bar, 850K and at equivalence ratios between 2.6 and 3.3. Methane, natural gas and ethylene (0.5 < φ < 2.5) failed to auto-ignite within the test section. Model predictions were found to agree with the ethylene measurements, although improved qualification of the experimental boundary conditions is required in order to better understand the dependence of auto-ignition delay on the physical characteristics of the flow rig. The chemical kinetic model used in this study was compared with existing ‘low temperature’ measurements and correlations for methane and natural gas and was found to be in good agreement.

The design of a premix duct is heavily reliant upon knowledge of the auto-ignition delay time for a given fuel-air mixture and operating range. An experimental investigation into the auto-ignition characteristics of alternative gaseous and liquid fuels has been carried out and the results compared against those for the conventional fuels. Gaseous fuels derived from sources such as bio-gas and refinery wastes were successfully compared against datum fuels of natural gas and methane — whilst among the liquid fuels — 100% biodiesel (consistent with BS EN 14214) and grade 2 diesel were tested and compared. The experimental results were obtained using a flow rig equipped with a generic premix duct and operated at GT-relevant conditions. For the gaseous fuels no ignition events were detected within the maximum test section residence time of 130ms. Therefore, a kinetic scheme previously validated at GT relevant conditions was used to evaluate effects of temperature, pressure and equivalence ratios for the gaseous fuels under investigation. In general gaseous fuels were found to have long auto-ignition delay times at temperatures < 900K and no differences were found between grade 2 diesel and 100% biodiesel. Therefore, the results under this study would be useful to the operators and designers of DLE systems when evaluating potential replacements for the current traditional fuels.

No abstract