ABSTRACT

As discussed already in Chapter 1, RANS methods can predict the flow quantities of interest to hydraulic engineers with sufficient accuracy for many relatively simple flows, that is when the flow is mainly unidirectional and largely attached as in channel-type flows. They still are the main workhorse of hydraulic engineers and allow practical flow calculations at reasonable cost, and not only at laboratory scale but also in field situations with high Reynolds numbers. However, RANS models have been found to be not sufficiently adequate for complex situations where largescale anisotropic structures are important and govern the flow behaviour and where unsteady effects are of interest such as unsteady forces on structures and situations with fluid-structure interaction. In such cases LES has been found to be clearly superior, but it is also much more expensive as LES will always be three-dimensional and time-dependent, and near walls the resolution requirements are such that the number of grid points necessary increases roughly with Re1.8 for a truly wall-resolving LES (Chapman 1979). Hence such LES are not affordable for real situations with large Reynolds numbers as they usually occur in practice.