Reduced-order modelling and quantifying local entropy production with triple decomposition in a low-pressure turbine

The precise quantification of energy losses induced by multiscale flow structures in turbomachinery remains challenging, particularly under periodic wake disturbances. One of the bottleneck problems, which is frequently encountered in critical conditions for turbomachinery, lies at recognizing the losses associated with the large-scale perturbations or coherent structures in turbulent flows. This study addresses the critical need to distinguish irreversible losses from large-scale coherent structures versus random turbulence in a low-pressure turbine. A framework integrating triple decomposition of entropy production with quadruple Proper Orthogonal Decomposition was developed, enabling spatiotemporal separation of mean flow, wake-driven coherent motions, and small-scale turbulence. In order to validate the ability of the method in quantifying the irreversibility of wake flows, a low-pressure turbine cascade with incoming periodic wakes was numerically analyzed with simulations. For what concerns the quantitative estimation of the entropy production by mean flow, coherent flow and random turbulent flow are obtained. The entropy production by mean flow accounts for almost 99% of the total energy. The main energy loss induced by the coherent structure are found near 90% axial length and suction side of blade, driven by the tip leakage vortex and passage vortex. The energy loss induced by small-scale turbulent flow is similar with that of coherent structure, except from a more significant impact by the incoming wakes at 30% axial length of blade downstream leading edge. The methodology advances conventional Reynolds-averaged entropy analysis by resolving scale-specific loss mechanisms, and establishes a paradigm for optimizing turbine designs against multiscale flow irreversibility.