Chapter 3 : Reservoir models - KU Leuven

the same parameters can be specified as for the impervious areas and as an extra
possibility the infiltration can be modelled using the Horton equation as described in
paragraph 3.2.4. The decay factors for surface wetting (equation 3.5) and drying
(equation 3.6) may be chosen differently. The depression storage on the impervious
areas can be emptied during dry weather using a constant evaporation. The possible
evaporation is strongly influenced by other climatic parameters as for instance
the temperature. However, the actual evaporation from impervious areas varies much
less, because of the limited amounts of water available in the depression storage
(figure 3.24). The evaporation has only a small effect on the losses and does not
influence the prediction of overflows directly. It only slightly influences the antecedent
conditions. Therefore, it is acceptable to use a constant evaporation for emission
predictions at combined sewer overflows. For the pervious areas the depression storage
can also be emptied by infiltration.
In the reservoir modelling system Glas, hourly rainfall was used which already
incorporated some smoothing. Furthermore, a time shift of one or more hours (multiple
of the time step) could be applied on the throughflow. This is sufficient if no better
accuracy than hourly based results is necessary and for sewer systems which have
concentration times of maximum a few hours. Larger concentration times seldom
occured, because in earlier times only small systems were modelled with a single
reservoir. Since 10 minute rainfall data are available now, it is possible to predict the
combined sewer overflows with higher accuracy. In order to obtain the smoothing of
the hydrograph and the time delay of the peak flow, it is necessary to add an extra
smoothing module in the reservoir modelling system. This smoothing is applied to the
runoff.
The smoothing module contains an averaging of the runoff over the concentration time.
In this way the peak shift between inflow and outflow are well determined and part of
the smoothing due to the flow in the system is already incorporated. This smoothing
is necessary, because a single reservoir model has an instantaneous (exponential)
response. This means that the peak shift cannot be predicted well in the reservoir
model, when no extra module for this is included. For the same reason (i.e. exponential
response) a single linear reservoir is not a good tool for this smoothing. If the
concentration time is used as a variable parameter (as a (hyperbolic) function of the
runoff), discontinuities may occur for low concentration times using only multiples of
the 10 minute time step. Therefore the concentration time is incorporated with a higher
accuracy of 1 minute, however without reducing the simulation time step. This also
allows a more accurate smoothing for small systems using a constant concentration
time. The use of the relationship between peak shift and maximum runoff as described
in paragraph 3.2.2 can be generalised for the smoothing of all runoff discharges (not
just only the peaks) in order to obtain continuous smoothing. This is physically
acceptable and has so far led to good results. However, already good results are often
obtained with a constant concentration time which is equal to the peak shift between
rainfall and outflow for the (composite) storm that will just lead to an overflow event.
3.28
The influence of rainfall and model simplification on combined sewer system design