Keywords: water distribution networks, leakage, pressure, minimum night flow, EPANET, Arc/View.

This paper introduces a methodology for leakage modelling in water distribution networks, by combining three software codes. At first, a computer code is developed to determine the components of real losses using the annual water balance method. It should be mentioned that only an estimated value can be obtained for real losses by this method. Then by applying the minimum night flow (MNF) method, the values of inflow and average pressure of the district metered area (DMA) are measured. The model calculates the leakage flow at the MNF time and the average daily and yearly leakage rates. The fixed and variable area discharge (FAVAD) and BABE concepts are used to evaluate leakage [1]. According to the field studies, the components of real losses (i.e. background losses, reported and unreported bursts, reservoir leakage and overflow, etc.) are also determined by the program. This information is useful for leakage management purposes.

In addition, the leakage model evaluates performance indicators of leakage using the IWA methodology [2]. These indices help the decision makers for better understanding of the network infrastructure conditions and choosing the optimum leakage management scheme. The values of unavoidable, current and economic annual real losses are evaluated in this regard. Then the infrastructure leakage index (ILI) and the feasible amount of leak reduction are determined. Leak location and detection schemes together with pressure management are common procedures for reducing the leakage.

At the second stage, a new methodology is introduced which can be performed by the EPANET hydraulic software to calculate nodal leakage values based on the measured inflow and network leakage results. Dividing the nodal consumption into two parts, volumetric and pressure dependent, the total network inflow is distributed to each node. In this procedure, nodal leakage flows are determined based on nodal pressures and calibrated nodal coefficients, while nodal volumetric parts of the consumption remains constant. Afterwards the pipe leakage rates are calculated.

At the final stage, all the network information including the nodal and pipe leakage data are exported to a GIS model (Arc/View). Using all features of the GIS model such as maps, data bases and network analysis options, several analyses can be performed in the GIS environment. For example the leakage risk map including the pipes with critical leakage rate, together with the burst density, pipe ages and maximum pressure maps can be obtained and used by decision makers.

After obtaining a realistic view from the network leakage and infrastructure situation, a number of scenarios such as leak detection, pressure management (PM), and pipe replacement can be selected to improve the network performance. A pressure management scheme which has a rapid and perfect effect on the leakage can be taken as the first choice. Following the pressure reduction, the procedure of step two is repeated and new nodal and pipe leakage flows are obtained. As an advantage, the leakage rates are reduced after pressure management. Therefore, the total consumption will be reduced. Furthermore, the optimum period to repeat leak detection plans can be determined based on this methodology.

Adopting the developed integrated model for leakage studies in an Iranian city, all the components of real losses were determined. The results showed high rates of 4638.38 and 1693010 (m³) for daily and yearly leakage, respectively. Also, a high value of ILI (14.95) and a low rate of economic efficiency (13.37%) illustrated that the network conditions were unacceptable. These indices highlighted the urgent need for leakage management schemes in this network. Nodal leakage was calculated by the EPANET and the highest amounts of nodal and pipe leakage were identified in the range of 0.445-1.18 and 0.512-1.523 (l/s), respectively. Two scenarios were considered for leakage management. At first, after the pressure management plan, the total yearly leakage is reduced to 584330 (m³), which means 34% reduction by installation of 6 PRVs across the network. Secondly, using the information produced by the model for real losses, including (background losses and reported and unreported bursts) and consideration of three periods of 6, 12 and 24 months for active leakage control, one year period was found to be economically suitable for leak detection which can reduce the real losses by 9%.

1

Lambert, A., "Pressure Management / Leakage Relationships: Theory, Concepts and Practical Applications", in "Proceedings of Minimizing Leakage in Water Supply / Distribution Systems", IQPC Seminar, London, April, 1997.

2

Farley, M. and Trow, S., "Losses in Water Distribution Networks", IWA Publishing, 2003.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £135 +P&P)