SINOPSIS

The proper orthogonal decomposition (POD) method is applied to analyze the database obtained from the direct numerical simulation (DNS) of supersonic plane mixing layers. The effect of different forms of the inner products in the POD method is investigated. It is observed that the mean flow contributes to a predominant part of the total flow energy, and the energy spectrum of the turbulence fluctuations covers a wide range of POD modes. The patterns of leading (high energy) POD modes reveal that the flow structures exhibit spanwise counter rotating rolls, as well as oblique vortices. These flow patterns are insensitive to the velocity of the observer. As the convective Mach number increases, the energy spectrum becomes wider, the leading POD modes contain more complicated structures, and the flow becomes more chaotic.