Separation in overflow spillways: a computational analysis

Abstract

Separation in overflow spillways may occur when the hydraulic head is considerably greater than the design head. In this study, results obtained in a separation that occurred in a physical model were used for a numerical analysis of the phenomenon. A three-dimensional computational programme was used and the results using none or one of six different turbulence models were compared to the experiment. The results indicate that the renormalization group (RNG) model has a better capacity to simulate the separation zone; therefore, this turbulence model is recommended. Once separation still occurred when no turbulence model was used, the use of non-appropriate models may conceal the phenomenon. This is a computational indication that viscous effects have no major part in this separation condition, with the spillway profile being the most decisive aspect. When RNG was the applied model for several different hydraulic heads, good agreement in terms of discharge coefficient between numerical and physical model was obtained.

Keywords:

• Computational fluid dynamics
• flow separation
• overflow spillway
• RANS models
• three- dimensional models
• turbulence model