Abstract
Two representative Low-Reynolds number models from two most widely used groups of two-equation turbulence models, together with their respective original standard models, have been put to detailed examination based on comparisons with DNS and experimental data on two-dimensional open channel flows. It has been found that recent fine adjustments to reproduce the DNS data in near-wall low-Reynolds number regions do not necessarily improve prediction of general flows with free surface in complex geometry. Appropriate treatment of the free-surface effects is more important. The low-Reynolds number version of the k-ω model showed the best predictive accuracy in the present study.
