The flow in a hydraulic turbine draft tube is a challenging industrial application for CFD because is characterized by a turbulent flow with different flow phenomena, e.g. unsteadiness, flow separation, swirling flow and strong adverse pressure gradient. Thus, its simulation is complex and time-consuming high computational capacities are required Additionally, adequate turbulence modeling is needed to predict such flows accurately. The objective of the present work is to investigate the accuracy of the Renormalization Group Theory (RNG) k-c turbulence model on the draft tube flow close to the best efficiency of the turbine. The effect of the discerization schemes on these models has been studied As the flow is significantlyaffected by the walls, two different grid concentrations near to the wall (y + I and y + 50) were evaluated The numerical results are compared to detailed experimental results at different section and discussed.