A simplified modellling approach for storm water flow optimisation
During a rainstorm, urban drainage systems channel the storm water to locations where
the water can be treated. Many urban drainage systems in the UK are quite old and were
originally created to cope with lower population levels. As the rate of urbanisation
increase and more surfaces become impermeable, drainage systems re inundated with
increasing amounts of water. Over time, urban drainage systems require significant
alterations to allow them to channel the increasing levels of storm water flows.
A model of the urban drainage system can be created to simulate the effects of storm
water in the system when different types of rainstorms are applied to it. Such models
can be used to predict the consequences of any alterations to the system. With limited
public funding available to improve existing systems, optimisation procedures offer a
possible method for identifying the best solution for the least cost. Simulating existing
urban drainage models is time consuming, especially with the amount of separate
simulations that are required by most optimisation procedures.
The aim of this study is the development of a simplified model for an urban drainage
system that is primarily designed for speed of calculation. Adapted from a simplified
pollution control model, the Packet Approach allows for the rapid evaluation of an
urban drainage system model under different rainfall events. Coupled with a Genetic
Algorithm optimisation method, the packet approach is easily applied to a basic model
to reduce flooding caused by an inefficient drainage network.
The scale of the Packet Approach is expanded so that it can be applied to a commercialsized
drainage system. The expanded version of the Packet Approach, called FastNett,
simulates commercial models and allows for a comparison of flow rate and flood
volume to those predicted by alternative commercial software. Once a reasonable
comparison is achieved, the model is optimised by FastNett, which dramatically reduces
the level of flooding for a minimal cost and in a reasonable calculation time. At the end,
several suggestions re given as to how FastNett could be adapted to overcome any of its
natural limitations and also to cope with more complicated commercial networks.