Environmental Engineering Reference
In-Depth Information
be reflected in the behaviour of the model. It is
widely recognized that adequately representing
the sediment impacts of hydraulic structures is
a challenge in practically all sediment models.
9 Optimizing run parameters such as choice of
layer thickness, critical flow for sediment entrain-
ment and method of cross-section updating - how
these user-defined parameters are set can have a
major impact on model outcomes.
Experiencedmodellers recognize many of these
issues and the pitfalls that lurk behind them and
they are adept at dealing with them too. They can
do this because they have first-hand knowledge of
how to address the issues and avoid the pitfalls,
based on how simulations have been finessed in
the past despite software limitations and weak-
nesses in process knowledge, such as how in-
channel sediment transport is affected when flow
spills over bank. Unfortunately, the guidance
available in the literature on dealing with these
issues is currently limited and improved dissem-
ination of both research findings and experience
gained during practical applications could help
improve the reliability and accuracy of 1-D sedi-
ment models that are built and run by less expe-
rienced modellers. For example, 20 years ago,
Samuels (1990) provided clear guidance on the
minimum spacing necessary to properly represent
flood hydrodynamics in a 1-D model, but up to
now no similar guidance has been published for
sediment modelling.
2 Selecting the period and duration to bemodelled
and deciding how to represent the flow - ideally,
continuous simulation is preferred, but for long
model runs it may be necessary to keep run times
manageable bymodelling only the sediment trans-
porting events.
3 Identifying the cross-section spacing needed to
represent sediment processes adequately - cross-
sections must be spaced more closely in a sedi-
ment model than is permissible when modelling
the flow of water alone.
4 Setting the sediment in flow to the modelled
reach - often there will be few or nomeasured data
and yet this boundary condition is important to
how the model will run. Usually, if no data are
available, the sediment input is set equal to the
transport capacity at the first computational
section, so that this boundary is stable.
5 Deciding how to handle sediment grading - a
decision must be made on whether it is necessary
tomodel sediment transport by size fraction and, if
it is, the range and number of size classes to be
used, how many samples must be collected to
characterize the size distribution sufficiently
well, and how much complexity to use in the
model (this ranges from using a different size
distribution for every cross-section to using a sin-
gle size distribution to represent the bed through-
out the modelled reach).
6 Selecting the sediment transport equation -
based on the scale and steepness of the water-
course, the sediment size distribution and the
model complexity, it is essential to select a sedi-
ment transport equation appropriate to the
application.
7 Identifying the simplifying changes necessary
to an existing hydrodynamic model required for it
to support the sediment module - for example,
successful simulation of in-bank sediment trans-
port may require reducing extended cross-sections
or using spills to removefloodflows to out-of-bank
storage.
8 Working out how best to represent in-stream
structures that impact sediment dynamics - sed-
iment traps, bridges and culverts and out-of-bank
flows around these structures disrupt connectiv-
ity in the sediment transfer system and this must
Limitations to 1-D Sediment Modelling
Having dealt with the issues underlying 1-D sed-
iment modelling, there remain limits to its appli-
cability that cannot be worked around easily or
simply. These include:
1 Three-dimensional effects of sediment trans-
port such as local scour at bridges are not ac-
counted for.
2 Sediment calculations for compound channels
and overbankflowsmay be less accurate due to the
lateral averaging of flow parameters.
3 Layers of sediment are assumed to be homoge-
neous across thewidth so that lateral sorting, such
as occurs at meander bends, is not represented.
