The role of water distribution networks in water supply

A water supply system comprises a raw water source, a treatment or production plant, and a distribution system. To achieve its main objective – i.e., to provide consumers with water at required pressures and flows – all three system components should perform at an optimum level. The raw water source should have sufficient yield, the treatment plant processes should ensure throughput that can satisfy demand, and the distribution system should be capable of delivering the water. If any component functions inadequately, the system as a whole might underperform. Often, if consumers do not receive the required water supply, it is believed that the problem lies with inadequate production. While this may be true, another factor that can seriously affect service levels is the quality of the distribution system.

The water supply system of Kampala, Uganda can be used as an example. After augmenting production to 160,000 m³/d, which exceeded demand at that time, consumers still frequently complained about water scarcity (NWSC 2007). This indicates that the cause of the system's underperformance is not necessarily production, but might be the distribution system. Due to population growth, water demand in Kampala is now (in 2018) 200,000 m³/d, which has further exacerbated the situation. This is not helped by an inadequate network whose pipes are poorly sized and have not received attention in 50 years. Kampala is not the only area that faces water shortages, but is among many urban areas that have pressurized supply systems with intermittent flows (Zyoud 2003; Khatri & Lee & Schwab 2005; Nyende-Byakika 2006; Vairavamoorthy 2007; Biswas & Seetharam 2008; Rosenberg 2008; Vairavamoorthy et al. 2008), which are characterized by low pressures and flows.

Water distribution systems comprise, essentially, reservoirs, pipes, pumps and valves. These need to be properly designed and optimized so that they can function adequately, delivering the required water volumes to consumers (Nyende-Byakika et al. 2010). In order to meet regulatory requirements and customer expectations, water utilities are feeling a growing need to improve explanations of the movement, etc, of water in their distribution systems (Kritpiphat et al. 1998; Rossman 2000). If understanding of network behavior under different conditions could be obtained and the impact of these conditions established, networks could be managed better and better customer service offered.

The objective of this paper is to demonstrate the significance and influence that water distribution networks can have on water supply, with particular reference to achieving the desired pressures and flows. The Rubaga subsystem of the Kampala water distribution network was selected as it is a typical example.

The methodology used for this research is similar to that reported by Nyende-Byakika et al. (2012). A network model of the Rubaga subsystem was developed using the EPANET2 engine (Rossman 2000) and scenarios were used to assess its performance. Inputs to the model included nodal demands and elevations, and pipe lengths, sizes and friction factors, as well as the hydraulic head of the reservoir. The model outputs of interest included pressures at nodes, as well as head-losses and velocities across pipes. Figure 1 shows the modelled subsystem, indicating the elevations of the nodes and the pipe diameters.

The model was tested and calibrated to field-observed values for a range of operating conditions, to evaluate its ability to represent actual situations.

In order to meet the objectives, the model was tested under two broad scenarios – observation of network performance using different pipe sizes, and, separately, the impact of location, height and reservoir sizing. The principal model outputs desired were flows, velocities and pressures under different scenarios, to demonstrate the impact of particular network statuses on the delivery of pressures and flows required by customers.

Figure 2 shows the variation of pressure at node 2 (Figure 1) with the size of the connecting pipe when satisfying a demand of 6.8 l/s. The node connects the Rubaga tank to the subsystem. Figure 3 shows head-loss variation with pipe diameter. In this case, a diameter less than or equal to 200 mm could not serve the subsystem because it would yield negative pressures, as the delivery flow required leads to high head-losses that would impede delivery in a small diameter pipe – i.e., less than 200 mm in this case. Pipe diameters between 300 and 900 mm yielded higher pressures for larger pipe sizes (Figure 2), in a non-linear relationship. This agrees with Equation (1), which shows that the greater the pipe diameter the lower the head-loss (yielding higher pressure).

For pipe diameters exceeding 900 mm, the pressure remains the same because the head-loss falls to negligible values with sufficiently large pipe diameters. At this stage, friction between moving water and the pipe walls is greatly reduced because water ceases to exert additional pressure against the walls when the cross-sectional area is large compared to the volume of water entering the pipe.

Figure 4 shows how high pressure results from low link head-loss. Head-losses represent losses in pressure and, thus, with high head-losses, pressure is low.

Figure 5 shows the variation of water velocity with pipe size (or cross-sectional area) for the same volumetric flow rate. The greater the pipe diameter the lower the flow velocity, in line with the principle of mass conservation.

Variation of pressure along a pipeline is demonstrated in Table 1. Pressure increases downstream along the pipe as the elevation falls. The lower the node elevation, the greater the height (static head) from which water is supplied, leading to higher pressures.

Other determinants of node pressures are the elevation at that point and at the supply tank. The greater the difference between the node and supply tank's elevations, the greater the nodal pressure. Table 2 shows the pressure variations at midnight at nodes 16 and 17, respectively at 1,184 and 1,175 masl (Figure 1). The supply tank elevation is modelled at different levels.

The paper shows that adequate water supply to consumers is achieved not only by increasing production, but also through maintaining an efficient distribution network. Head-loss can be controlled in pipelines by using larger diameter and/or smooth and new pipes, and by limiting pipe lengths. This can be done by connecting consumers to larger pipes with short runs to supply tanks in the network. It is also important to note that pressures increase with an increase in the static head between the supply point and the node.

It is very important to select optimally-sized pipes to deliver the highest pressures and flows, while avoiding both under- and over-sizing, which both increase capital and operating costs. It is also important to run a hydraulic simulation of the network to investigate how changes to one part or parameter might have an impact elsewhere. This arises from conflicting constraints, for example, while enlarging pipe diameters can increase pressures, it reduces flow velocity. Thus, network optimization is very important to obtain the best results.

It follows directly and in agreement with the principle of mass conservation, that, to further improve supply, there should be mechanisms by which water drawn from the system is controlled, such as by rationing through valve control.

The author declares that there is no conflict of interest regarding the publication of this paper.