Genetic algorithms and GIS data for decision making in planning water distribution networks

This thesis is concerned with the optimal design of Water Distribution Networks (WDNs). The design involves finding an acceptable trade-off between cost minimisation and the maximisation of numerous system benefits. The primary design problem involves cost-effective specification of a pipe network layout and pipe sizes in order to satisfy expected consumer water demands within required pressure limits. The design of a WDN has many variable parameters such as position and size of the water sources, position and the size of the pipes and position of the treatment plants. However, the layout is constrained by the location of existing facilities such as streets and buildings and other geographic features. The total costs may consist of the cost of network materials such as pipes, construction works and system operation and maintenance. The problem may be extended to consider the design of additional components, such as reservoirs, tanks, pumps and valves. Practical designs must also cater for the uncertainty of demand, the requirement of surplus capacity for future growth, and the hydraulic reliability of the system under different demand and potential failure conditions. The thesis reviews the literature related to water distribution networks, their design and optimisation. It then presents a Genetic Algorithm (GA) formulation to assist in developing the design of a water distribution network. The main aim of this research is to investigate the possibility of combining GAs and GIS in the design optimisation. A decision mechanism is developed which enables the model to reach a meaningful solution and provide a practical design technique for WDNs. The aim is also to provide an experimental analysis of the combined GA and decision mechanism to solve the problem in hand and to assess the robustness of these techniques when applied to different instances. An initial prototype model is presented for the design of a WDN which is used to determine the necessary features of the 'final' model. These features include the world in which the model will be built, the design of the fitness function, chromosome representation, and GA operators. The research mainly concluded that the initial model prototype was useful to determine the necessary features and to produce the final model which enables a variety of necessary factors to be explicitly included in the design of WDNs. This initial model suggested that the final model should include the decision mechanism, which is a matter of policy management and hydraulics, and hydraulic principles which allowed to compare the behaviour of different parameters and to simulate the functioning of the network under different scenarios. Water allocation and distribution policies can be applied according to the importance of the demand area and the ability of the system to deliver sufficient water amounts. These policies link essential hydraulic and institutional relationships as well as water uses and users and allocation decision-making process. It was also found that the representation of the world layout is important. The world is described in GIS in terms of models that define the concepts and procedures needed to translate real-world features into data. The important aspects in the chromosome representation are the node positions, the links. In this case, a chromosome must contain the three-dimensional node coordinates, the connection between nodes, the head required to pump the water. The best model parameters were extracted to be used in real-life situations. The result of tests on an example world demonstrated that the model was successful, and the potential exists for the use of this formulation in more complex and real-world scenarios.EThOS - Electronic Theses Online ServiceGBUnited Kingdo