Abstract
The article presents a data processing methodology to identify the noise sources in low-pressure turbine (LPT) stages and their relative importance. Scale adaptive simulation (SAS) has been carried out on a geometry reproducing the midspan blade section of an entire LPT stage. The proper orthogonal decomposition (POD) is applied to data matrices constructed with the velocity and the pressure fields in order to distinctly extract the coherent structures responsible for velocity fluctuations and pressure waves (i.e., POD modes from the corresponding kernel), their temporal evolution and energies. The cross-correlation matrices of the pressure modes and the velocity modes reveal the degree of coupling between pressure and velocity oscillations, thus highlighting the modes linked to the same flow phenomena. The results show that the periodic vortex shedding at the stator trailing edge emits strong pressure waves, and the upstream-propagating waves reflect against the adjacent stator suction surface. The POD modes related to the rotor–stator interaction show peak amplitudes at the blade passing frequency and its harmonics and their shapes mimic the potential interaction in the stator domain and the migration of viscous wakes in the rotor domain. The POD modes with lower energy content correspond to the stochastic turbulent structures originated from the turbulent boundary layers as well as those carried by the vane and blade wakes. The biggest noise sources of the current flow fields are identified to be the von Karman vortex street downstream the stator trailing edge and the rotor–stator interaction.
