and quasi-steady water-flow equations, has shown that the uncoupling is not a serious obstacle
to successful simulation.
Another distinguishing feature of numerical bed-evolution models is the representation of
sediment sorting and bed-surface armoring. Alluvial sediments are rarely of uniform grain size.
A broad range of sizes are represented, from gravels and coarse sands down to fine silt and
clay in varying proportions. Finer particles are preferentially entrained into the flow as erosion
occurs, so that the material remaining on the bed contains a progressively higher proportion of
coarser material. This so-called sorting process tends to increase the mean bed-sediment size
as degradation occurs, thus affecting the sediment-transport rate, river regime (existence of
ripples and dunes), and flow resistance through both particle roughness and bed-form effects.
If the original bed material contains a high enough proportion of large, non-moveable materials
(coarse gravel, cobbles, and small boulders), an interlocking armor layer may form on the
surface, arresting further degradation. These processes are qualitatively reversed during
deposition, but become even more difficult to quantify. Some models attempt to simulate
sorting effects on bed evolution; others ignore them completely. Thus another important
distinguishing feature of computer-based models is the degree to which they incorporate
sorting and armoring effects.
Numerical modeling of alluvial river flows has become very popular in the 1990s because of the
advancement of computer technology. However, the number of computer-based, alluvial
riverbed prediction models that are readily available for application to prototype cases seems to
be quite small. Many of the available models have been developed for specific rivers under
particular flow and alluvial river bed conditions, and many of them are, to some extent, well
tuned or calibrated only for those particular rivers.
The following assessment of selected models is made for two different groups: short-term
models and long-term models. The short-term models are best suited to compute changes in
alluvial river bed level during a relatively short time period. Long-term models employ simpler
implementations of steady-state flow equations, and thus are suited for long-term prediction of
river bed level for multiple-flood events over multiple years. However, it should be recognized
that the short-term models can also be applied for long-term prediction if variable time steps
are employed. In that case, a shorter time step is used for highly unsteady flows and a longer
time step is used otherwise.
BRI-STARS MODEL (The Bridge Stream Tube Model for Alluvial River Simulation). The
BRI-STAR (Molinas 2000) model developed for the Federal Highway Administration is a
semi-two-dimensional model capable of computing alluvial scour/deposition through subcritical,
supercritical, and a combination of both flow conditions involving hydraulic jumps. Both energy
and momentum equations are used so the water surface profile computation can be carried
out through combinations of subcritical and supercritical flows without interruption. The stream
tube concept is used in a semi-two-dimensional way, which allows the lateral and longitudinal
variation of hydraulic conditions as well as sediment activity at various cross sections along the
study reach. The sediment continuity equation and sediment transport capacity equations are
used for sediment routing computations to simulate the general scour in the river bed elevation.
The sediment routing is performed for each size fraction to account for the bed composition
changes and the bed armoring processes. The minimum rate of energy dissipation theory is
used for decisions as to whether channel adjustments taking place at a given cross section.
BRI-STARS is capable of simulating channel widening/narrowing phenomenon.