Prediction of Pressure Oscillations in a Premixed Swirl Combustor Flow and Comparison to Measurements

The most common and reliable technique used for flame stabilization of industrial combustors with high thermal loads is the application of strongly swirling flows. In addition to stabilization, these swirl flames offer the possibility to influence emission characteristics by simply changing the swirl intensity or the type of swirl generation. Despite of these major advantages, swirling flows tend to evolve flow instabilities, that considerably constitute a significant source of noise. In general, noise generation is substantially enhanced, when such a swirling flow is employed for flames. Thus, the minimization of the resulting noise emissions under conservation of the benefit of high ignition stability is one major design challenge for the development of modern swirl stabilized combustion devices. The present investigation makes an attempt to determine mechanisms and processes to influence the noise generation of flames with underlying swirling flows. Therefore, a new burner has been designed, that offers the possibility to vary geometrical parameters as well as the type of swirl generation, typically applied in industrial devices. Experimental data has been determined for the isothermal flow as well as swirl flames with 3-D-LDV-diagnostics comprising the components of long-time averaged mean and rms-velocities as well as spectrally resolved velocity fluctuations for all components. The noise emission data was acquired with microphone probes resulting in sound pressure levels outside the zone of perceptible fluid flow. In addition to the experiments, numerical simulations using RANS and LES have been carried out for the isothermal case. The results of the measurements show a distinct rise of the sound pressure level, obtained by changing the test setup from the isothermal into the flame configuration as well as by varying geometrical parameters, which is also resembled by the LES simulation results. Additionally, a physical model has been developed from experiments and verified by the LES simulation, that explains the formation of coherent flow structures and allows to separate their contribution to the overall noise emission from ordinary turbulent noise sources.