21

Locating Leaks in Water Distribution

Systems Using Network Modeling

Paul F. Boulos, Trent Schade and Chris Baxter

Water distribution systems can experience high levels of leakage resulting in

major financial, supply and pressure losses. Locating and repairing system
leaks can drastically reduce the amount of water that is lost, as well as
reduce the costs for obtaining, treating and pressurizing water distribution
systems to meet current and future demands. This chapter describes an
efficient step-testing network modeling approach that solves the leakage
detection problem using a direct application of network modeling and field
testing. The technique involves bracketing the test area with excessive
leakage into a tight branched network with a flow meter installed on its input
main. Working from the valve furthest away from the flow meter, the size of
the area is systematically reduced by closing valves to cut off different pipe
sections in succession (so that less and less of the test area is supplied
through the meter), at the same time recording changes in flow rate at the
meter and comparing with model results. The sequence of closing valves is
followed working backward towards the flow meter until the meter is
reached (when the flow becomes zero). A disproportionate change in flow
discrepancy between two successive steps indicates a leak in the section of
pipe that was last shut off. The sequence is repeated by opening valves in
reverse order. The method can effectively narrow down leaks to specific
pipe segments of the distribution system. It is normally carried out at night
before the morning high demand to minimize supply interruption and
inconvenience to customers. An example application is used to illustrate the

Boulos, P., T. Schade and C.W. Baxter. 2008. "Locating Leaks in Water Distribution Systems Using Network
Modeling." Journal of Water Management Modeling R228-21. doi: 10.14796/JWMM.R228-21.
© CHI 2008 www.chijournal.org ISSN: 2292-6062 (Formerly in Reliable Modeling of Urban Water Systems.
ISBN: 978-0-9808853-0-9)

351
352 Locating Leaks in Water Distribution Systems using Network Modeling

proposed approach. The method should prove useful to any water utility
attempting to locate excessive pipe leaks in distribution systems and
conserve such a precious natural resource as water. It is modeling
complexity through simplicity.

21.1 Introduction
One of the major contributors to water loss is underground leaks in water
distribution systems. These systems can experience high levels of leakage
resulting in major financial, supply and pressure losses. Leakage occurs in
different components of the system including transmission and distribution
mains, service connection lines, valves, joints, and fire hydrants. It can
originate from many sources such as a the deterioration of aging pipes and
fittings, material defect, changes in water pressure (water hammer), high
population density, heavy traffic volumes, movement of above ground
pipelines, aggressive soil conditions, and corrosion (AWWA, 1999).
Excessive leakage can also cause contaminant intrusion events, which can
lead to detrimental or fatal water quality episodes. The EPA estimates that
water utilities in the US will need to spend US$6 billion per year over the
next twenty years to rehabilitate failing water distribution pipes (Lansey and
Boulos, 2005; Boulos et al, 2006).
Locating and repairing system leaks can drastically reduce the amount of
water that is lost, the implementation of drastic conservation policies during
droughts, adverse risk of contamination, and water outage events, as well as
reduce the costs for obtaining, treating, and pressurizing water supplies.
Other benefits for water utilities and their customers include more efficient
use of existing supplies, improved water conservation measures,
environmental quality, and system operational efficiency, integrity,
reliability, performance and fire fighting capability, as well as extended
useful life of existing facilities, increased services to new developments, and
delayed treatment plant capacity expansion and construction of new sources
of water supply (Lauber, 1977).
System (water) audits and leak detection programs can help utilities
reduce water losses (AWWA, 1999). Water audits involve detailed
accounting of the distribution system inflows and outflows. They provide an
overall picture of the distribution system efficiency and water losses and
identify areas of the system experiencing excessive leakage. The overall goal
is to identify, quantify, and verify water and revenue losses. These audits do
not provide information about the precise location of leaks. Leak detection
Locating Leaks in Water Distribution Systems using Network Modeling 353

programs identify and prioritize the areas of high leakage using step-testing.
The principal objective of step-testing is to continuously isolate portions of
the distribution system where leakage is measured quantitatively (WHO,
2001). Localizing leaks can also be carried out using acoustic listening
devices (sounding). The exact locations of leaks can then be determined
using leak noise correlators (leak noise correlation).
Acoustic loggers are normally installed on pipe fittings and are used to
identify suspected leakage areas by listening for leak characteristics. By
recording and analyzing the intensity and consistency of noise, the loggers
are able to determine the likely presence of a leak. Noise is created by the
leak as it escapes from the pressurized pipe. Similar to the traditional sonic
equipment, the correlator relies upon the noise generated by a leak. The
main difference, however, is in how the leak noise is detected. A correlator
works by detecting the sound from the leak when it arrives at two sensors
(installed on pipe fittings) on the pipe, either side of the suspected leak
position. The difference in the arrival time of the leak noise at each sensor
(time delay), combined with knowledge of the distance between the two
sensors and the velocity of the sound in the pipe, enables the correlator to
calculate the leak position. Normally, acoustic equipment is effective for
metal pipes as the leak signals transmit for relatively long distances but
could be problematic for plastic piping as the signals only transmit for only
very short distances. In general (Hunaidi et al, 2000), leak noise correlators
are more efficient and more accurate than listening devices. Leaks can also
be located with non-acoustic technologies such as tracer gas, termography,
and ground-penetrating radar. The use of these techniques is, however, still
very limited and their effectiveness not well established (Hunaidi, 2000).
Step-testing is one of the most effective and practical tools of identifying
and quantifying leakage within specific areas (step areas) of the distribution
system. This is an activity whereby the step area is subdivided by the
systematic closing of valves during the period of minimum night flow and
recording the reduction in flow. Disproportionate drops in flow define the
pipes with suspected leakage. The method is particularly suited to
identifying areas of high leakage and to use on plastic pipes, where leak
noise is absorbed and conventional acoustic methods are less effective. It
also minimizes the use of detection required by sonic equipment and
correlators. Having localized the leaks using step-testing, the correlators can
be used to pinpoint the precise locations of the leaks along the pipes, and
repairs can then be carried out. The sequence of step testing and leak
detection and repair is continued until an acceptable level of leakage in each
step area is recorded.
354 Locating Leaks in Water Distribution Systems using Network Modeling

The main drawbacks of the traditional step-testing approach are the

requirement of detailed maps of the distribution system, showing all water
facilities (pipes, valves, pumps, tanks and reservoirs), and the need to
determine the proper sequences of valve closing to cut off different pipe
sections in succession. In addition, the method requires consumption to be
relatively stable and its efficacy is limited by the lack of database support
and user-friendly interfaces for data manipulation, hardcopy reporting
(i.e. detailed field implementation manual), and graphical output display.
These requirements suggest the need for an integrated network modeling
approach to step-testing.
This chapter presents a rigorous and efficient step-testing modeling
approach that overcomes these limitations. It combines hydraulic network
modeling with field testing. The network model contains all pertinent water
system facilities and is used to delineate the step area, provide the necessary
flow and consumption characteristics, and to compute the optimal patterns of
systematic closing of valves. In addition, the method is less sensitive to the
structure of the network and to the variation in demands. These capabilities
provide a consistent modeling environment to assist water utilities in
planning, designing and implementing cost-effective and reliable leak
detection programs. The method is illustrated by application to an example
water distribution system and conclusions are stated.

21.2 Methodology
The proposed step-test network modeling approach narrows down leaks to
specific pipe segments of the water distribution system. It involves network
modeling and field testing. Network modeling determines how to define and
subsequently subdivide the step area with the systematic closing of valves,
and computes the total flow into the step area associated with each distinct
valve closing operation. Field testing involves taking flow readings while
performing required valve operations (Figure 21.1). At each step, the flow
reading (at the flow meter) is taken and compared to the modeled value, and
the difference in flow rate is recorded. Any leakage isolated in a particular
pipe segment sequence is shown as a drop/change in the difference in flow
(flow discrepancy) between two successive steps. The flow discrepancy also
represents the suspected leakage in the associated pipe segment sequence.
Locating Leaks in Water Distribution Systems using Network Modeling 355

Figure 21.1 Field operation for step-testing.

The method involves bracketing an area with excessive leakage into a

tight zone (step area) with a flow meter installed on the input main to each
zone to record flow into the zone. The step area must be fed via only one
source (e.g. storage tank, pipe interconnection) and must not contain any
loops (tree or branched network). The step area thus becomes a tree-like
structure that is supplied through a single meter and isolated from the rest of
the system by closing all boundary and circulation valves. This can be
accomplished by:
· closing boundary valves to isolate the step area from the
system (only one source of supply)
· closing circulating valves to remove loops and create a tree
network
· placing a data logger to the flow meter
Figure 21.2 shows the schematic of a sample step area. The initial area is
depicted in Figure 21.2a while the final tree-like step area, obtained after
closing boundary and circulating valves, is shown in Figure 21.2b. It is also
important to identify all step valves that can be used during the test and all
other valves that should be excluded from operation (not to be used during
the test) such as boundary valves and inoperable (e.g. broken) valves.
Identification of the latter group of valves is important to avoid mistakenly
opening those valves during the test.
Working from the farthest valve (away from the meter), the size of the
zone is systematically reduced by closing valves to cut off different pipe
sections in succession (so that less and less of the test area is supplied
356 Locating Leaks in Water Distribution Systems using Network Modeling

through the meter), at the same time recording changes in flow rate at the
meter and comparing with modeled results. Each step must be shut long
enough to see the flow impact at the meter. The sequence of closing valves
is followed until the flow meter is reached (when the flow becomes zero). A
significant discrepancy in flow rate between two successive steps indicates a
leak in the section of pipe that was last shut off. The sequence is repeated by
opening valves in reverse order. The opening and closing of valves should
be performed slowly to avoid unwanted surges and breaks. Step-testing is
normally carried out at night before the morning high demand when the
consumption is lowest and relatively unchanged (static), and to minimize
supply interruption and inconvenience to customers. Flushing may be
required prior to step-testing a targeted area of the distribution system to
alleviate potential water quality problems resulting from valve operation
(WHO, 2001).

Figure 21.2 Sample step-test area definition.

Step-testing can be summarized as follows:

1. Step-test at night during period of low/slack demand;
2. Define step area with suspected high leakage;
3. Close all boundary valves (to establish a tight area);
4. Close all circulating valves to remove loops and create a
tree network;
Locating Leaks in Water Distribution Systems using Network Modeling 357

5. Attach data logger to the flow meter installed on the main

supplying the step-test area;
6. Start at the pipe/valve farthest from the flow meter;
7. Close step valves in succession such that less and less of
the step area is supplied via the flow meter and record
measured flow values;
8. Follow the sequence of closing valves (as determined by
the network model) back towards the meter, when the
reading is zero;
9. Keep each step long enough to notice a reading impact at
the meter; and
10. Reopen the step valves in reverse order.

It should be noted that depending on the topological structure of the step

area, more than one arrangement of step sequences could be possible. For
example in the sample step area shown in Figure 21.3, the following distinct
sequences are feasible:

(V8,V7,V6,V5,V4,V3,V2,V1), (V8,V7,V3,V2,V6,V5,V4,V1),
(V8,V7,V6,V5,V3,V2,V4,V1), (V3,V2,V8,V7,V6,V5,V4,V1),
(V3,V2,V6,V5,V8,V7,V4,V1), (V6,V5,V8,V7,V4,V3,V2,V1),
(V6,V5,V8,V7,V3,V2,V4,V1), (V6,V5,V3,V2,V8,V7,V4,V1).

The network model can identify all possible step sequence arrangements
and the desired sequence selected by field personnel based on optimal route.

Figure 21.3 Example of feasible step sequences.

358 Locating Leaks in Water Distribution Systems using Network Modeling

21.3 Illustrative Example

A worked out example is presented here to illustrate the calculation steps of
the network modeling approach for leak detection. The step-test area is
shown in Figure 21.4. The test area comprises 6 pipe sections and 6 demand
nodes. Two leaks (of one flow unit each) are assumed for pipes P2 and P4.

Figure 21.4 Illustrative step-test modeling example.

Figure 21.5 Flow step results towards the meter for sample test area.
Locating Leaks in Water Distribution Systems using Network Modeling 359

Figure 21.5 gives the flow results for each step sequence back towards
the flow meter. Moving from right to left in Figure 21.5, field crews will
sequentially shut valves. With each valve closure, the flow rate is reduced at
the monitoring point on the left side of the figure. The valve closing
operations at valve V2 and valve V4 result in two distinct flow
discrepancies, indicating the presence of a leak in pipes P2 and P4.

Figure 21.6 Complete step-step flow results for sample test area.

Figure 21.6 shows the flow results for the entire step-test. It should be
noted that the symmetry along the Y-axis will only be valid when the
demand loadings (and associated patterns) and operating conditions over the
step-test period remain unchanged (static). However, the step discrepancies
(the additional flow from leaks) will not change depending on the demand
loading changes. This observation is key for this method, as the redundancy
of measurements provides a stable and efficient experimental design that can
be reproduced under any demand conditions. The flow calculations for the
example network are summarized in Table 21.1.
360 Locating Leaks in Water Distribution Systems using Network Modeling

Table 21.1. Step-test calculations for sample test area.

Step Time Valve Flow Flow Flow Step Comment

Seq. (a.m.) Operation Measured Modeled Difference Discrepancy
0 12:10 -- 14 units 12 units 2 units --
1 12:20 V1 12 units 10 units 2 units 0 unit No leak
2 12:30 V2 9 units 8 units 1 unit 1 unit 1 unit
leak
3 12:40 V3 7 units 6 units 1 unit 0 unit No leak
4 12:50 V4 4 units 4 units 0 unit 1 unit 1 unit
leak
5 1:00 V5 2 units 2 units 0 unit 0 unit No leak
6 1:10 V6 0 unit 0 unit 0 unit 0 unit No leak

21.4 Leak Location

Once leakage has been localized to specific pipes within a step area, leak
noise correlation can be effectively used to identify precisely where the leak
is located along each pipe. In this method, two sensors are installed on either
side (usually hydrant or valve) of the suspected leak position. The location
of the leak ( X ) can then be computed using a simple algebraic relationship
(based on the principle of correlation) between the difference in the arrival
time of the leak noise at each sensor ( Δt ), which is measured from the
cross-correlation of the leak signals, the distance between sensors ( L ), and
the propagation of sound waves in the pipe ( V ) as:

X = ( L − (VΔt )) / 2 (21.1)

This approach is illustrated in Figure 21.7.

Figure 21.7 Principle of correlation for leak detection.

Locating Leaks in Water Distribution Systems using Network Modeling 361

21.5 Conclusions
Step-testing is the process of localizing leakage into specific pipe segments
of the distribution system for subsequent replacement or repair. This chapter
has presented an efficient approach to optimally identifying pipes with
suspected leaks by combining field testing with network modeling. The leak
in a pipe segment is determined by analyzing the rate of change in the
discrepancy between field measured and modeled flow values. The method
is well suited to bracket high leakage areas in the system and is applicable to
any pipe materials. In addition, it is less sensitive to the topological structure
of the network and to the variation in consumption than the traditional step-
test field approach.
The method can be further enhanced with linkage to a geographic
information system (GIS). The GIS can be used to store, locate, manage and
display all pertinent water system facilities and produce comprehensive
maps of the step areas. The data from the GIS environment are fed into the
network model that produces the optimal step-testing sequences, which in
turn can be evaluated by the GIS to provide utility personnel with a detailed
field step-testing implementation manual detailing the proper sequences of
valve operations. The resulting geospatial step-test modeling approach will
help to effectively communicate the schedule/progress information to field
personnel, because they will be able to see in detail the temporal sequences
of valve operations alongside the modeling results (Boulos, 2006). It is
modeling complexity through simplicity.

References
AWWA (1999). “Water Audits and Leak Detection.” Manual of Water Supply Practices
No. M36, 2nd Edition. American Water Works Association, Denver, CO, 99 pages.
Boulos, P. F. (2006). “InfoWater LDM – A Comprehensive GIS-Centric Network
Modeling Program for Leak Detection.” MWH Soft Pub., Pasadena, CA.
Boulos, P. F., Lansey, K.E. and Karney, B.W. (2006). “Comprehensive Water
Distribution Systems Analysis Handbook for Engineers and Planners.” Second
edition, MWH Soft Pub., Pasadena, CA, 660 pages.
Hunaidi, O., Chu, W., Wang, A. and Guan, W. (2000). “Detecting Leaks in Plastic
Pipes.” Journal American Water Works Association, 92(2), pp. 82-94.
Hunaidi, O. (2000). “Detecting Leaks in Water Distribution Pipes.” Construction
Technology Update No. 40, Institute for Research and Construction, National
Research Council of Canada, pp. 1-6.
Lansey, K.E., and Boulos, P.F. (2005). “Comprehensive Handbook on Water Quality
Analysis for Distribution Systems.” MWH Soft Pub., Pasadena, CA, 448 pages.
362 Locating Leaks in Water Distribution Systems using Network Modeling

Lauber, C.E. (1997). “Leak Detection Cost-Effective and Beneficial”. Journal American
Water Works Association, 89(7), pp. 10.
World Health Organization (2001). “Leakage Management and Control – A Best Practice
Training Manual.” World Health Organization.