2014 年 2 巻 1 号 p. 159-167
Momentum transfer of open-channel flows in the lateral direction is important in sediment transport, flood control, environmental issues, etc. in rivers. The lateral transfer of the streamwise momentum is known to be caused by three different mechanisms: a cross-sectional secondary current due to turbulence anisotropy (secondary current of the second kind), turbulence mixing with shear instability (K–H instability), and mass transfer from the rough-bottomed to the smooth-bottomed lane due to flow redistribution. Furthermore, Vermaas et al. (2011)1) have shown experimentally that the relative contribution of each mechanism to the momentum transfer is closely affected by depth. Thus, to elucidate the fundamental characteristics of the lateral momentum transfer, we performed three-dimensional (3-D) computations of shallow open-channel flows in two parallel lanes with beds of different roughness.
We simulated the momentum transfer on roughness transition using a 3-D RANS (Reynolds Averaged Navier Stokes) type κ–ε turbulence model, which has significantly greater computational efficiency than DNS (Direct Numerical Simulation) or LES (Large Eddy Simulation). Two types of computations, focusing on either the well-developed or the developing process, were performed separately. The present numerical results show that although the linear κ–ε model completely fails to capture the fundamental characteristics of the flows, a second-order non-linear κ–ε model could estimate excellently each effect on the momentum transfer, including the dependence on flow depth.