Sedimentary facies in the distal parts of deep-marine lobes can diverge significantly from those predicted by classical turbidite models, and sedimentological processes in these environments are poorly understood. This gap may be bridged using outcrop studies and theoretical models. In the Skoorsteenberg Fm., a downstream transition from thickly-bedded turbidite sandstones to argillaceous, internally layered hybrid beds is observed. The hybrid beds have a characteristic stratigraphic and spatial distribution, being associated with bed successions which generally coarsen- and thicken-upwards reflecting deposition on the fringes of lobes in a dominantly progradational system. Using a detailed characterisation of bed types, including grain size, grain fabric and mineralogical analyses, a process-model for flow evolution is developed. This is explored using a numerical suspension capacity model for radially spreading and decelerating turbidity currents. The new model shows how decelerating sediment suspensions can reach a critical suspension capacity threshold beyond which grains are not supported by fluid turbulence. Sand and silt particles, settling together with flocculated clay, may form low yield-strength cohesive flows; development of these higher concentration lower boundary layer flows inhibits transfer of turbulent kinetic energy into the upper parts of the flow ultimately resulting in catastrophic loss of turbulence and collapse of the upper part of the flow. Advection distances of the now transitional to laminar flow are relatively long (several km) suggesting relatively slow dewatering (several hours) of the low yield strength flows. The catastrophic loss of turbulence accounts for the presence of such beds in other fine-grained systems without invoking external controls or large-scale flow partitioning, and also explains the abrupt pinch-out of all divisions of these sandstones. Estimation of the point of flow transformation is a useful tool in the prediction of heterogeneity distribution in subsurface systems.