Spectral Separation of the Turbofan Engine Coherent Combustion Noise Component

The study of combustion noise from turbofan engines has become more important as other noise sources like the airframe, the fan and the jet are reduced. However, modern engines are based on new materials and a longer history of experience in turbofan engine manufacturing, design, analysis, and optimization which has lead to engines with reduced coherent combustor noise. In addition, the jet mixing noise from near the end of the potential core masks the low frequency core noise. This makes it necessary to update the coherent output power spectrum method so that the coherent combustion noise spectrum can be determined in spite of its low level. The core noise component of a dual spool turbofan engine (Honeywell TECH977) for engine power settings of 48 percent, 54 percent, and 60 percent were separated by the use of a coherence function. A method has been developed to help identify combustion noise coherence using an aligned and unaligned coherence technique which enables the validation of low levels of coherence as being due to core noise by identifying the coherence noise floor. A statistical procedure is also used to establish this threshold level. The use of both methods provides a high confidence level for the coherence function values calculated. A source location technique based on adjusting the time delay between the combustor pressure sensor signal and the far field microphone signal to maximize the coherence and remove as much variation of the phase angle with frequency as possible was used. These techniques make it possible to quantify the weak coherent core noise in the aft quadrant instead of dismissing it as negligible. While adjusting the time delay to maximize the coherence and minimize the cross spectrum phase angle variation with frequency, the discovery was made that for the 130 o microphone a 90.027 ms time shift worked best for the frequency band from 0-200 Hz while a 86.975 ms time shift worked best for the frequency band from 200-400 Hz. Since the 0-200 Hz signal took more time to travel the same distance, it is slower than the 200-400 Hz signal. This suggests the 0-200 Hz coherent cross spectral density band is partly due to indirect combustion noise attributed to hot spots interacting with the turbine. The net travel time of the indirect combustion noise signal from the combustor to the far field is increased since the travel velocity of the hot spots to the turbine and in the turbine is the flow velocity which is some small fraction of the speed of sound. The indirect combustion noise signal does not travel with the speed of an acoustic wave until it interacts with the turbine. The signal in the 200-400 Hz frequency band is attributed mostly to direct combustion noise which has the travel time of the acoustic wave in the combustor and turbine. Beyond the turbine both direct and indirect pressure signals travel at the speed of an acoustic wave to the far field. Consequently, this source separation method identifies in the turbofan engine ”direct” combustion noise and ”indirect” combustion noise due to ”hot spots” convecting through the turbine.