A critical evaluation of the Baldwin-lomax (Bl) turbulence model for shock/shear layer interactions, reversed flow, and curved, asymmetric wakes is made. No general definition for reference line is available for wakes, and difficulties exist for length scale prediction in complex flows. An entropy envelope for shear layers, and the locus of maximum entropy to define the wake centerline is proposed. The range of the Bl model is limited to the entropy envelope. This provides all relevant modeling data, and allows general application of existing reversed flow corrections. The total enhancements are flow adaptive and form the Dynamic Bl model. This robust model is more accurate in complex boundary layers and wakes. The Dynamic Bl model is applied to a supersonic fan cascade at the design incidence. Sharp differences in turbulent viscosities were seen between the the original, Baseline Bl, and Dynamic Bl models. Only slight differences exist in the overall cascade solutions. This includes loss factors which were produced by different mechanisms. The Baseline BL model predicted separation on the SS surface and larger standing vorticies off the TE. The Dynamic Bl model predicted attached boundary layers, smaller standing vorticies off the TE, but uniformly higher skin friction. The shock structure in the cascade may reduce the flow field dependence on specific viscosity profile characteristics, so these may be less important than overall turbulence levels. / Master of Science