Cross-correlation and coherence functions are used to look for periodic acoustic components in turbofan engine combustor time histories, to investigate direct and indirect combustion noise source separation based on signal propagation time delays, and to provide information on combustor acoustics. This investigation uses a combustor pressure sensor and a far-field microphone at 130 to study the change in propagation time to the far field of the direct and indirect combustion noise signal using a generalized cross-correlation function. Results are presented as a function of the cutoff frequency of the low-pass filter used to create the generalized cross-correlation function and engine operating condition. The filtering procedure used produces no phase distortion. As the low-pass-filter frequency is decreased, the travel time increases. The indirect combustion noise signal travels more slowly, because the entropy in the combustor moves with the flow, which has a low velocity. The indirect combustion noise signal travels at acoustic velocities after reaching the turbine and being converted into an acoustic signal. The direct combustion noise is always propagating at acoustic velocities. This is in agreement with previous investigations of delay times using a cross-spectrum phase-angle method with unfiltered signals that found indirect combustion noise to be in the 0-200 Hz frequency range and the direct combustion noise to be in the 200-400 Hz frequency range. Similar results obtained using the cross-spectrum phase-angle method with filtered signals are also shown. These results show that the low-pass filtering can be used with the cross-correlation function to separate this type of dependent source and confirm the cross-spectrum results. Although the results are based on a set of static engine tests conducted for one specific dual-spool turbofan engine configuration, they may lead to a better idea about the acoustics in the combustor turbine-tailpipe system and may help develop and validate improved reduced-order physics-based methods for predicting turbofan engine core noise.
NomenclatureB e = resolution bandwidth, Hz, 1=T d r s =N 2 Hz c c = combustor-region speed of sound, m=s c o = ambient speed of sound, m=s D = propagation time delay or lag, s f c = upper frequency limit, 1=2t r s =2, Hz (32,768 Hz) f L = cutoff frequency of low-pass filter L c = combustor-region length, m L y = number of frequencies, f c =f N=2 (16,384) M c = combustor-region Mach number N = segment length, number of data points per segment (32,768) n o = number of overlapped data segments/blocks n s = number of disjoint (independent) data segments/ blocks, B e T total 128 P I = confidence-interval percent value used to calculate coherence threshold < = gas constant, 8314 J=k mol K R x = autocorrelation function R xy = cross-correlation function r = microphone radial location, m (30.48 m) r s = sample rate, samples/s (65,536) T c = temperature in combustor, K T d i = record length of segment i, N=r s , 0.5 s T total = total record length, s ...