Compact and non-compact analytical solutions of the subsonic operating point of the entropy wave generator experiment are compared with detailed numerical results obtained by large Eddy simulations. Two energy deposition methods are presented to account for the experimental ignition sequence and geometry: a single-block deposition as previously used and a delayed deposition that reproduces the experimental protocle closely. The unknown inlet acoustic reflection coefficient is assumed to be fully reflective to be more physically consistent with the actual experimental setup. The time delay between the activation of the heating modules must be considered to retrieve the temperature signal measured at the vibrometer and pressure signals at the microphones. Moreover, pressure signals extracted from the large Eddy simulations in the outlet duct using the delayed ignition model clearly reproduce the experimental signals better than the analytical models. An additional simulation with actual temperature fluctuations directly injected at the inlet of the computational domain clearly shows that the pressure fluctuations produced by the acceleration of the hot slug yields indirect noise almost entirely. Finally, the entropy spot is shown to be distorted when convecting through the turbulent flow in the entropy wave generator nozzle. Its amplitude decreases and its shape is dispersed, but hardly any dissipation occurs. The distortion appears to be negligible through the nozzle and become important only when convected over a long distance in the downstream duct. As the dominant frequencies of the entropy wave generator entropy forcing are very low, the effects of dispersion by the mean flow are however weak.
Numerical investigation of combustion noise: The Entropy Wave Generator