Download as pdf or txt
Download as pdf or txt
You are on page 1of 18

Structure and Infrastructure Engineering

Maintenance, Management, Life-Cycle Design and Performance

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/nsie20

Developing a framework for leakage target setting


in water distribution networks from an economic
perspective

Iman Moslehi, Mohammadreza Jalili-Ghazizadeh & Ehsan Yousefi-Khoshqalb

To cite this article: Iman Moslehi, Mohammadreza Jalili-Ghazizadeh & Ehsan Yousefi-Khoshqalb
(2021) Developing a framework for leakage target setting in water distribution networks from
an economic perspective, Structure and Infrastructure Engineering, 17:6, 821-837, DOI:
10.1080/15732479.2020.1777568

To link to this article: https://doi.org/10.1080/15732479.2020.1777568

Published online: 11 Aug 2020.

Submit your article to this journal

Article views: 82

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=nsie20
STRUCTURE AND INFRASTRUCTURE ENGINEERING
2021, VOL. 17, NO. 6, 821–837
https://doi.org/10.1080/15732479.2020.1777568

Developing a framework for leakage target setting in water distribution


networks from an economic perspective
Iman Moslehi, Mohammadreza Jalili-Ghazizadeh and Ehsan Yousefi-Khoshqalb
Department of Civil, Water and Environmental Engineering, Shahid Beheshti University, Tehran, Iran

ABSTRACT ARTICLE HISTORY


Leakage is one of the main sources of water loss in Water Distribution Networks (WDNs). Although a Received 18 May 2019
water utility intends to reduce the leakage to the minimum possible level, considering the high leak- Revised 15 January 2020
age reduction costs, an economic perspective of leakage control should be taken into account. In this Accepted 13 February 2020
paper, a field-data based methodology for estimating the Short-Run Economic Level of Leakage (SR-
KEYWORDS
ELL) with respect to Active Leakage Control (ALC) activities is proposed. This methodology uses meas- Active leakage control;
ured pressure and flows data for the estimation of different leakage levels and relevant performance economic level of leakage;
indicators. The objective is to identify the target level of leakage in terms of the SR-ELL, using a mar- infrastructure condition;
ginal cost approach. Results of the sensitivity analysis are also presented to investigate the factors operating pressure; water
influencing SRELL. The methodology is applied to a large zone of WDN in the city of Mashhad, Iran. distribution networks
For this WDN, based on the current ALC policy and operating pressure, SRELL is estimated to be
27 m3/service connection/year. The results reveal that the SRELL is significantly affected by the operat-
ing pressure and infrastructure condition. The proposed methodology can be used by water utilities
for having a clear strategy on their leakage control policies.

1. Introduction Leakage management may be defined as collective activ-


ities and techniques, performed by water utilities to minim-
Leakage in water distribution networks (WDNs) remains a
ize real losses economically; it generally involves assessment,
serious concern for water utilities worldwide, as a significant
detection, and control of system leakage (American Water
amount of treated and transferred water in WDNs is lost
Works Association [AWWA], 2016; Kanakoudis, Tsitsifli,
due to leakage. The volume of leakage varies widely from 3-
Cerk, et al., 2015; Kanakoudis, Tsitsifli, & Zouboulis, 2015).
7% of supplied water in well-maintained networks in devel-
Leakage management practices and techniques can be cate-
oped countries to more than 50% in developing countries
(Gupta & Kulat, 2018; Kanakoudis & Tsitsifli, 2010; gorized into three main repetitive stages: (1) leakage moni-
Kanakoudis, Tsitsifli, & Demetriou, 2016; Kanakoudis, toring and assessment methods, (2) leakage detection,
Tsitsifli, Samaras, & Zouboulis, 2013; Puust, Kapelan, Savic, localization and pinpointing and (3) leakage control strat-
& Koppel, 2010). Asian Development Bank (2010) reported egies and methods (Gupta & Kulat, 2018; Kanakoudis &
that in Asian countries, more than 21 billion cubic metres Muhammetoglu, 2014; Puust et al., 2010; Tsitsifli et al.,
of treated water is lost through water leakage in WDNs, 2017). Leakage monitoring and assessment methods involve
costing water utilities $6.5 billion US (Frauendorfer and quantifying the volume of water lost in a distribu-
Liemberger, 2010). tion network.
Saving half of this amount provides enough water for at There are three primary assessment methods: (1) top-
least 150 million people. In Iran, water leakage from urban down water balance, (2) component analysis based on the
WDNs has been reported to be as high as 27% (National concept of burst and background estimates (BABE) and (3)
Water and Wastewater Engineering Company (NWWEC), Bottom-up approach using minimum night flow (MNF)
2018). In addition to financial losses, leakage may cause analysis (AL-Washali, Sharma, & Kennedy, 2016;
energy losses, service interruption, and water contamination Kanakoudis, Tsitsifli, Kouziakis, & Lappos, 2015; Thornton,
by the intrusion of polluted water (Mutikanga, Sharma, & Sturm, & Kunkel, 2008). MNF analysis is a powerful tool
Vairavamoorthy, 2013; Puust et al., 2010; Tsitsifli & and the most widely used method in practice for leakage
Kanakoudis, 2018). Therefore, leakage has become a crucial assessment (Amoatey, Minke, & Steinmetz, 2018; Gupta &
issue to be encountered by water utilities due to adverse and Kulat, 2018; Mazzolani et al., 2017). Moreover, it is a field-
costly consequences. For all these reasons, one of the main based method which is performed through measuring flow
responsibilities of the water utilities is to efficiently control and often pressure in individual zones or district metered
and reduce leakage in WDNs. areas (DMAs) (AL-Washali et al., 2016; Mutikanga et al.,

CONTACT Mohammadreza Jalili-Ghazizadeh m_jalili@sbu.ac.ir Department of Civil, Water and Environmental Engineering, Shahid Beheshti University,
Tehran 5357116589, Iran
ß 2020 Informa UK Limited, trading as Taylor & Francis Group
822 I. MOSLEHI ET AL.

2013). MNF is the measured flow supplied to a discrete Gonelas & Kanakoudis, 2015). Moreover, the sustainable
zone during the night period with minimum demand, usu- economic level of leakage (SELL) can be estimated by add-
ally occurring between 2:00 and 4:00 a.m. Water flow during ing social and environmental costs and benefits (external-
this time of the day is at a minimum because consumption ities) to the internal costs of the utility (Ashton & Hope,
is generally at its lowest, and therefore, leakage forms the 2001; Howarth, 1998; Lambert et al., 2015; Smout, Kayaga,
largest share of the total flow (Amoatey, Minke, & & Munoz-Trochez, 2010).
Steinmetz, 2014; Thornton et al., 2008). ALC is an effective proactive policy that a water utility
Leakage management strategies and intervention methods implements for reducing unreported leakage subject to avail-
focus on effective controlling and reducing leakage to as able operating budget (AWWA, 2016). The SR-ELL for an
great an extent as possible (AWWA, 2016; Gupta & Kulat, ALC activity has been reached when the SRMC of water
2018; Puust et al., 2010). These strategies include active exceeds the marginal cost of ALC work. There are two dif-
leakage control (ALC), pressure management (PM), passive ferent ways for determining SR-ELL of ALC activity. The
leakage control (PLC), sectorization and asset management IWA Water Loss Task Force developed a simple and prac-
(Farley & Liemberger, 2005; Haider et al., 2019; Kanakoudis tical methodology in terms of the natural rate of rise (NRR)
& Gonelas, 2016). The final choice among different leakage of unreported leakage to calculate the economic intervention
reduction strategies or a combination of activities should be frequency (EIF), and the SR-ELL of an ALC policy which
selected depending on economic consideration and the char- referred to as rate of rise (RR) method (Lambert &
acteristics of the system. In practice, water utilities aim at Fantozzi, 2005; Lambert & Lalonde, 2005). EIF can be calcu-
developing a leakage management strategy to establish the lated using three system-specific parameters including the
economic balance among all intervention methods that offer marginal cost of water, rate of rise of unreported leakage
the most cost-effective way of reducing water losses (Fanner and intervention cost (Alkasseh, Adlan, Abustan, & Hanif,
et al., 2007; Kanakoudis & Gonelas, 2015; Lambert, Merks, 2015; Kanakoudis & Gonelas, 2016; Lambert & Fantozzi,
& Trow, 2015; Pearson & Trow, 2005). Additionally, leakage 2005; Lambert & Lalonde, 2005; Munoz-Trochez, Smout, &
target setting in terms of economic level of leakage (ELL) is Kayaga, 2011).
the most important aspect of any leakage management strat- This method is only suitable for utilities that have not
egy (Islam & Babel, 2013; Trow & Farley, 2004). been operating under steady-state conditions. This means
ELL is defined as the optimal level of leakage at which that the water utility has been undertaking little or no ALC
the cost of reducing leakage equals the benefit gained from effort, and therefore, there is a backlog of unreported leaks
further leakage reductions. At this leakage level, the mar- in the network (Fanner et al., 2007). Moreover, the RR
ginal cost of leakage reduction is equal to the marginal cost method estimates only the economic unreported leakage
of water saved (Fanner et al., 2007; Gonelas & Kanakoudis, corresponding to the estimated EIF. Therefore, to calculate
2016; Pearson & Trow, 2005; Tripartite Group, 2002). There SR-ELL, other components of leakage (i.e., reported leaks
are two time periods for ELL calculations, which differ and background leakage (BL)) need to be predicted based
mainly from the time required to perform interventions, on BABE component analysis. Furthermore, predictions and
namely short-run and long-run with different marginal costs calculations of SR-ELL using the RR method are subject to
(Fanner et al., 2007; Kanakoudis & Gonelas, 2016). The considerable uncertainty, specifically associated with the
short-run marginal cost (SRMC) is defined when only oper- NRR estimation of unreported leakages. The value of NRR
ating costs are included in the ELL calculations. Based on is remarkably system-specific depending on operating pres-
this definition, the short-run economic level of leakage (SR- sures, seasonality, infrastructure condition, and operational
ELL) can be determined by comparing the marginal cost of procedures (Bettin, Rogers, & Serrani, 2014; Thornton et al.,
short-run leakage control measures such as ALC, PLC, and 2008). Hence, an approximate assessment of the NRR of
some PM schemes with the short-run marginal cost of water unreported leakages is often used to estimate the EIF, which
(SR-MCW) (i.e., only production and distribution operating can be introduced as an element of uncertainty into the
costs such as power and chemicals). RR method.
The long-run marginal cost (LRMC) comprises both The second methodology is the marginal cost approach
operating and capital costs in ELL calculations (Pearson & on the basis of the total annual cost of ALC intervention
Trow, 2005; Julian Thornton et al., 2008). There are some and water lost due to leakage (Lim, Savic, & Kapelan, 2015;
leakage control activities involving capital expenditure on Tripartite Group, 2002). In this method, the total cost curve
infrastructures like network rehabilitation, additional PM can be derived by adding the ALC costs to the cost of water
and sectorization (Chisakuta et al., 2011; Lambert & lost. The SR-ELL occurs at the minimum point of this
McKenzie, 2002; Trow & Farley, 2004). In these cases, the curve. This method provides reliable results when the net-
long-run economic level of leakage (LR-ELL) is required, work is in a steady-state condition. The water utility is not
which evaluates the investment decisions within the full life- projected to have a supply-demand deficit over the plan-
cycle assessment. The net present value (NPV) analysis can ning horizon.
be used to calculate the costs and savings of each leakage There have been a few studies to determine SR-ELL for
reduction option. The option having the lowest NPV is the ALC activity using the marginal cost approach (Islam &
least cost strategy, and the corresponding leakage defines Babel, 2013; Lim, Savic, & Kapelan, 2015). This approach is
the LR-ELL (Creaco & Walski, 2017; Fanner et al., 2007; data-intensive and comparatively complex, requiring a
STRUCTURE AND INFRASTRUCTURE ENGINEERING 823

collection of sufficient-field data, which is a challenging task points of the zone. Next, recorded high-frequency flow and
to accomplish for water utilities (Fanner & Lambert, 2009). pressure data (usually every 15 min), network characteristics
The SR-ELL is also affected by several influential factors, (e.g., pipe diameter, pipe length, number of connections), cost
including PLC and ALC policies adopted by the utility, net- data related to ALC activity, and the marginal cost of water are
work operating pressures, marginal cost of water, marginal collected. In the third step, the current level of leakage is esti-
cost of ALC, and infrastructure condition. Previous studies mated using MNF analysis, which is based on night flow meas-
focused primarily on the ELL estimation, and a few have urements and leakage-pressure relationships (AL-Washali et al.,
investigated the influential factors on the SR-ELL. Fantozzi 2016). In this step, the unavoidable annual real losses (UARL),
and Lambert (2007) developed a practical way to include as well as the actual or target background leakage (TBL), are
PM’s effects in calculations of SR-ELL using the RR method. calculated. Then, SR-ELL, one of the main components of a
They concluded that the influences of pressure on all three successful leakage management policy, is estimated. Eventually,
components of leakage, on frequencies of pipe breaks and sensitivity analysis is conducted to investigate how the inde-
economic ALC, should be taken into account in SR-ELL pendent variables may affect the SR-ELL.
calculations.
Fanner and Lambert (2009) also presented a step-by-step
approach to incorporate confidence limits for assessing SR-ELL 2.1. Data collection
by means of RR method with PM, ALC, and leak run-time Sufficient and reliable field data measurements, including
options. Islam and Babel (2013) investigated two influential fac- pressure and flow, are essential for calculating the current
tors, including operating pressure and the marginal cost of leakage levels and leakage performance indicators (i.e., ILI
water on estimating SR-ELL. They also developed two nomo- and ICF). For each zone or sector, pressure should be meas-
graphs on the basis of leakage volume and the two influencing ured at the AZP. This is a physical location where the pres-
factors for quick estimation of SR-ELL. However, other sure variations are considered to be representative of the
essential factors like the marginal cost of ALC and the infra- average pressure (Pav) in the zone (ILMSS Ltd, 2013;
structure condition are not considered in their study. Renaud et al., 2015). The average zone night pressure
The main objective of this study is to propose a method- (AZNP) can also be measured, which is the average pressure
ology for leakage target setting in terms of the economic level at the AZP during MNF (Fanner et al., 2007). In this paper,
of leakage regarding the ALC policy in WDNs. Different leak- a systematic approach developed by International Leakage
age levels and leakage performance indicators are estimated Management Support Services (ILMSS), is applied to deter-
using collected field data in the proposed methodology. The mine the AZP in an individual zone for night flow analysis
need for identifying adequate points for collecting sufficient (ILMSS Ltd, 2013).
pressure and flow data is also highlighted. The current level of This approach comprises the following steps: First, the
leakage is calculated by conducting night flow analysis and weighted average ground level (WAGL) is calculated using
assessing appropriate night-day factors (NDF). A practical pro- GIS tools for an appropriate infrastructure parameter (i.e.,
cedure is developed to estimate minimum achievable leakage mains length, service connections, and hydrants). Second, a
level with unlimited ALC resources using infrastructure condi- convenient measurement point is selected near the centre of
tion of the network and different considerations of the water the zone with the ground level close to the WAGL, prefer-
utility. The methodology uses estimated leakage levels and cost ably on mid-sized mains. Finally, the instantaneous pressure
data to develop the ALC cost curve and identifies SR-ELL. values are measured continuously with a regular time step
Moreover, the influence of different variables on SR-ELL is (usually every 15 min).
investigated by conducting sensitivity analysis. This method is For estimate the leakage using the MNF analysis, flow
also applied to a large zone of WDN of Mashhad city in Iran rates should be measured at main points of the zone (i.e., at
to evaluate its practical application. the inlet and outlet points as well as the large consumers if
present) with a regular time step, usually shorter than
15 minutes (Alkasseh, Adlan, Abustan, Aziz, & Hanif, 2013;
2. Methodology
Farley & Trow, 2007). The collected flow data need to be
The present methodology uses various field data and net- validated, cleansed, and normalized to a regular time step.
work characteristics for estimating SR-ELL based on the Data validation and cleaning entail outlier detection and
marginal cost approach concerning ALC policy in WDNs. identification of missing or corrupt data, which occur due
This methodology comprises the following main steps: (1) to metering and telemetry problems. In the data normaliza-
determination of appropriate measurement points, (2) col- tion process, the desired time step is defined to obtain data
lection of required field data, (3) calculation of different with a regular time step (Loureiro et al., 2016).
leakage levels, (4) estimation of SR-ELL, and (5) sensitivity The concept of Fixed and Variable Area Discharge
analysis (Moslehi, 2020). (FAVAD) explains that the leak flow rate varies with pres-
Figure 1 depicts the flowchart of the proposed methodology. sure changes to the N1 exponent (Lambert, 2001; Thornton
Firstly, the average zone point (AZP) for measuring instantan- & Lambert, 2005; Thornton et al., 2008). The values of N1
eous pressure values is identified using a geographic informa- exponent can be assessed from three different sources of
tion system (GIS) tools and network characteristics. data: 1) field tests on WDNs and empirical predictive equa-
Simultaneously, instant flow rates are measured at the inlet tions, 2) laboratory tests and 3) theoretical modelling
824 I. MOSLEHI ET AL.

Figure 1. Flowchart of the proposed methodology for determining SR-ELL in WDNs.

(Lambert, Fantozzi, & Shepherd, 2017; Thornton & ALC cost curve. The costs are required to maintain leakage
Lambert, 2005; van Zyl, Lambert, & Collins, 2017). Field at a given steady-state level and increase at lower leakage
tests are conducted by a pressure step test or N1 step test levels (Pearson & Trow, 2005). It is widely agreed that the
that is an important part of leakage assessment methods shape of the ALC cost curve is a hyperbola that is asymp-
(Fanner et al., 2007). The test is conducted by regulating totic to a minimum leakage level (Fanner et al., 2007;
inlet pressures on an existing pressure reducing valve (PRV) Pearson & Trow, 2005; Tripartite Group, 2002).
in a series of steps during the night-time period. For modelling the steady-state ALC costs for different
Subsequently, the pressure drop at the AZP and the reduc- leakage levels, it is crucial to split the current costs into the
tion inflow into the zone are recorded at the start of the test fixed and variable parts. Leakage monitoring costs are
and each reduction step (AWWA, 2016; Thornton counted as a part of the fixed cost, as they do not vary with
et al., 2008). the leakage level. However, leakage detection and repair
costs can be considered variable cost elements because they
depend on the amount of ALC effort (UK Water Industry
2.2. Cost data
Research (UKWIR), 2011). These costs would increase for a
The first requirement in the marginal cost approach is to lower level of leakage as the company intensifies ALC activ-
estimate the current steady-state ALC costs for modeling ities and vice versa. It should be noted that leakage repair
STRUCTURE AND INFRASTRUCTURE ENGINEERING 825

costs consist of PLC and ALC repairs. Obviously, only the The NDF should then be obtained to convert night leak-
costs which are entirely attributable to ALC repairs are age rate to daily leakage rate by involving pressure changes
needed for modelling ALC cost curve. (AL-Washali et al., 2016; Amoatey et al., 2014; Lambert &
Best practice guidance recommends that the current Taylor, 2010). The NDF accounts for the influence of pres-
steady-state leakage level and the associated ALC costs sure variations during the day on leakage flow rates. The
should be defined for a recent period with no demand value NDF can be calculated using pressure measurements
restrictions and no changes in leakage control policy at the AZP and an appropriate N1 exponent value as fol-
(UKWIR, 2011). The fixed and variable costs of the steady- lows:
state ALC can be estimated as follows: i¼23 
X N i¼23  N
AZPi, iþ1 1 X AZPi, iþ1 1
CCAL ¼ CCALFix þ CCALVar (1) NDF ¼ ¼ (3)
i¼0
AZNP i¼0
AZP3  4
CCALVar ¼ NLD
where N1 ¼ the exponent of pressure-leakage relationship
 ½ðCLm  Lm þ CNC  NC Þ (FAVAD N1 exponent), AZP and AZNP ¼ the average
þ ðCBp þ CRp ÞðNumM þ NumC Þ (2) hourly and night pressure at the AZP, respectively.
The minimum level of real loss is defined as the lowest
where CCAL ¼ total annual costs of ALC effort (TIRR/year), technically achievable annual real loss, which could be
CCALFix ¼ annual fixed costs of ALC activity(TIRR/year), achieved at any particular operating pressure for well-man-
CCALVar ¼ annual variable costs of ALC activity (TIRR/year), aged infrastructure in good condition (Lambert, 2009;
NLD ¼ number of leakage detection efforts to maintain current Lambert and McKenzie, 2002). The IWA Water Loss Task
leakage level of the network, CLM ¼ unit cost of Leakage survey Force developed the basic equation for the calculation of
for mains (TIRR/km length of mains), CNC ¼ unit cost of leak- UARL. Recently, the following equation was recommended
age survey for service connections (TIRR/service connections), by (Lambert et al., 2014; Lambert, Koelbl, & Fuchs-
CBp Bp ¼ unit cost of the bonus payment to the contractor for Hanusch, 2014) to calculate UARL in m3/year:
detected leaks on mains and or service connections (TIRR/
detected leak), CRp ¼ unit cost of repaired leaks on mains and UARLðm3 =yearÞ ¼ ð6:57  Lm þ 0:292  NC þ 9:13  LP Þ
or service connections (TIRR/repaired leak), NumM and NumC  P  CP
¼ the number of detected and repaired leaks on mains and
(4)
service connections, respectively, Lm ¼ mains length (km), NC
¼ the number of service connections. where LP ¼ the total length of service connections from the
The next key parameter in the marginal cost approach is property line to the meter (Km), P ¼ the average operating-
the value of water lost, which includes the marginal cost of system pressure (meters), and CP ¼ pressure correction fac-
water production and distribution from existing water tor to consider the influence of pipe materials on UARL as
resources of the utility. This marginal cost is known as the pressure changes (Figure 2).
SR-MCW and usually comprises operating costs (Fanner In fact, the UARL consists of unavoidable background
et al., 2007). The SR-MCW can be calculated using the dif- leakage (UBL) plus losses from reported and unreported lea-
ference in operating cost in producing and distributing one kages. The UBL is the lower limit for BL, which is unavoid-
more or less unit of water in terms of power, chemicals and able for every zone, even newly commissioned (Fanner
possibly labor plus any volume-related costs of purchasing et al., 2007). The following equation calculates the UBL on
or withdrawing raw water from existing sources (Pearson & mains and service connections up to property line (Lambert,
Trow, 2005; Thornton et al., 2008). 2009):
 
AZNP 1:5
UBLðlitres=hourÞ ¼ ð20  Lm þ 1:25  NC Þ 
2.3. Estimation of the current level of leakage 50
The steady-state current leakage level in the proposed meth- (5)
odology is estimated using MNF analysis, which is the most The Infrastructure Leakage Index (ILI) is a performance
widely used method for determining leakage level in practice indicator quantifying how effectively a water utility controls
(Mazzolani et al., 2017; Mutikanga et al., 2013; Yousefi- the level of real loss by implementing leakage control meas-
Khoshqalb, 2019). The term MNF is defined as the lowest ures like PLC, asset management, and ALC at the current
flow supplied into a hydraulic network zone, which usually operating pressure (Lambert, 2002; Lambert et al., 2014).
occurs between 02:00 and 04:00 a.m. when most users are Mathematically, it is a dimensionless ratio between the
inactive (AL-Washali et al., 2016; Alkasseh et al., 2013; CARL and the UARL. It is important to note that the ILI is
Puust et al., 2010). Therefore, leakage accounts for a larger a pure technical performance indicator that does not con-
fraction of water flow during this time of the day. The vol- sider any economic considerations (Chisakuta et al., 2011;
ume of leakage can be estimated by subtracting the expected Liemberger, 2002).
legitimate night-time consumption (LNC) from the MNF. The general conditions of mains and service connections
The obtained value is called the net night flow (NNF), have a significant effect on leakage levels in WDNs. The
which estimates the leakage flow rate during the MNF effects of infrastructure condition on the level of BL are
period (Hamilton & Mckenzie, 2014). referred to as infrastructure condition factor (ICF), which is
826 I. MOSLEHI ET AL.

Figure 2. Provisional relationship between pressure and CP, for systems with different percentages of rigid pipes (Source: (Lambert, 2009)).

the ratio of the actual level of BL in a zone to the UBL. The there is no backlog of unreported leakage, except those
ICF determines the volume of BL under the current operat- occurring since the last survey (Fanner et al., 2007).
ing condition, which is the minimum achievable leakage The marginal cost of a leakage reduction strategy is an
level in a zone after all unreported leakages have been found important factor in SR-ELL calculations. For ALC activity, it
and fixed using conventional ALC methods (Fanner, et al., can be calculated using the ALC cost curve modelling,
2007; Farley & Trow, 2007). The ICF can be calculated which can provide estimates of ALC costs for maintaining
based on actual field measurements. leakage at different levels. The steady-state ALC cost curve
In this method, comprehensive leak detection and repair can be modelled based on the current leakage level, and the
is conducted in individual zones. After all, detectable leaks associated ALC costs to maintain that level, as well as an
have been removed; then, the remaining BL can be directly estimate of the background and passive leakage levels (UK
measured (AWWA, 2016; Fanner et al., 2007; Thornton Water Industry Research (UKWIR), 2013).
et al., 2008). This value can be taken as actual or target BL. The TBL is a minimum achievable level of leakage using
If the utility does not have sufficient leakage repair data, limitless ALC resources (AWWA, 2016; UKWIR, 2013). The
there is a practical method to estimate the ICF. It can be ALC costs are asymptotic to this point, which means that
taken by assuming the ICF to be approximately equal to the the leakage level may be approached to the TBL when the
ILI value when the target leakage levels have been achieved. marginal cost is ever-increasing. The second parameter to
define the ALC cost curve is the passive level of leakage
The target ILI can be identified using preliminary leakage
(PLL). Leakage level reaches PLL when the utility takes a
management target-setting guidelines based on water resour-
PLC policy (i.e., no ALC activity is performed) while
ces, operational and financial considerations of the water
responding only to leaks reported by the customer or utili-
utility (AWWA, 2016).
ty’s staff (UKWIR, 2011). The PLL can be calculated using
NNF data from the same period of MNF analysis. ALC cost
2.4. Economic analysis of leakage curve at the PLL falls to the fixed costs of leak-
age monitoring.
As previously mentioned, the proposed methodology deter- The last parameter for modelling cost curve is the
mines the SR-ELL with respect to ALC activity based on the steady-state ALC costs for the current leakage level, which is
marginal cost approach (Fanner et al., 2007; Lim et al., calculated using Equations (1) and (2). The steady-state
2015; Tripartite Group, 2002). In this approach, the mar- ALC cost curve is then modelled using the following equa-
ginal cost of ALC is compared with the marginal cost of tion (UKWIR, 2011):
water lost. The SR-ELL has been reached when the marginal  
L  Lb
cost of leakage control activity exceeds the marginal cost of C ¼ k  ln (6)
LP  Lb
water. Moreover, a steady-state condition for the network is
necessary to apply this approach in practice, which means where C ¼ variable element of ALC costs, L ¼ leakage level,
STRUCTURE AND INFRASTRUCTURE ENGINEERING 827

Figure 3. ALC cost curve, the cost of water lost and total cost curve for estimating SR-ELL.

Lb ¼ target/actual background leakage, Lp ¼ passive leakage Subsequently, the SR-ELL calculations need to be repeated
level while k ¼ coefficient determined by: until the modified ICF reaches the new target ILI.
C
k¼  a  (7)
La  Lb 2.5. Sensitivity analysis
ln Lp  Lb
Estimation of the ELL is subject to uncertainty, as deter-
Ca ¼ ALC costs at the current steady-state leakage level, mined using a variety of variables (Fanner & Lambert,
La ¼ current or actual level of leakage. 2009; Lim et al., 2015). Sensitivity analysis is carried out
The marginal cost of ALC is the additional cost of reduc- to investigate how the SR-ELL may be affected by
ing the current level of leakage to a lower level and depends changing independent variables in the proposed
on how much water loss control efforts have been made by methodology.
the water utility (Islam & Babel, 2013). The marginal cost of
ALC is obtained by the gradient of the ALC cost curve at a
point on the curve, which is represented by: 3. Case study
The proposed methodology is applied to an individual zone
dC k
¼ (8) of a WDN in Mashhad (Iran). This is the largest city in
dL L  Lb
north-eastern Iran, at approximately 1500 metres above the
Once the ALC cost curve has been generated, it can be sea level. Figure 4 illustrates zone J, which was selected for
compared to the cost of water lost at different levels of leak- the implementation of the proposed methodology. The
age that is generally a straight line with the slope of the WDN of zone J consists of two subzones: J1 and J2, with
marginal cost of water. The total cost curve can be derived the approximately 112.5 km mains length and 26566 service
when the ALC cost curve and the cost of water lost are connections with 99% customer metering. This zone is sup-
added together. Figure 3 shows a typical total cost curve, plied by gravity from three inlet points. General characteris-
which tends to infinity at the TBL. The SR-ELL is the leak- tics of the studied zone are presented in Table 1. The
age level at which the marginal cost of ALC activity will be predominant pipe materials for the zone J comprise of 58%
asbestos cement concrete (ACC), 22.7% polythene (PE) and
equal to the marginal cost of water, which can be found at
17.5% ductile iron (DI). Most of the mains have diameters
the point of the minimum total cost.
less than 250 mm (88.5%). Moreover, the vast majority of
Once the SR-ELL has been established, the following
the pipes are less than 10 years old.
comparison should be undertaken in order to cross-check of
the estimated ICF and correct it, if necessary. The economic
ILI is calculated with the ratio of estimated SR-ELL to the 4. Results and discussion
UARL as a new target ILI. The new target ILI should be
4.1. Calculation of average zone pressure
close to the former target ILI which is identified based on
the preliminary leakage management target-setting guide- As the GIS data of service connections are not available, the
lines for accepting the estimated ICF. Differently, the esti- WAGL is assessed by superimposing the layout of the WDN
mated ICF is considered equal to the new target ILI. and digital elevation model (DEM) of the zone. For each
828 I. MOSLEHI ET AL.

Figure 4. Representation of the Study area (Zone J) and J1 and J2 subzones of the WDN of Mashhad. FM ¼ flow meter, PL ¼ pressure logger. (Image # 2018
Digital Globe, # 2018 Google Earth; Map data # 2018 Google.).

contour band, the ground level at the mid-point of the band In the examined case, the Mashhad water and wastewater
is multiplied by the counted mains length within the band, company (MWWC) has conducted leak detection surveys
then totalled. Finally, the WAGL is calculated by dividing annually to detect leaks and breaks on mains and service con-
the total obtained value of the former step over the total nections in the last three years since Mar. 2015. Furthermore,
length of the pipes. The values of WAGLs for subzone J1 all leaks and bursts found have also been repaired. Thus, the
and J2 were found to be 1125.7 and 1115.5 metres, respect- network is operating under a steady-state condition. Moreover,
ively. Figure 4 presents the selected points as the AZPs with flow monitoring into zone J is performed continually to meas-
the ground levels close to the obtained WAGLs and near ure leakage. Unit costs for various components of variable
the centre of the subzones. Instantaneous pressure values ALC costs are summarized in Table 2 based on the data
were measured at the AZPs of the subzones every 15 min. obtained from the MWWC. Accordingly, the variable costs of
The average pressure (Pav) for zone J is then calculated as ALC activities (Ca) for the study area are estimated to be
the weighted average pressure of the two subzones where 52.45 thousand IR Rials (TIRR) (USD 1.249) per service con-
the weighting factor is the mains length. nections (SC) per year. Furthermore, the leakage monitoring
Flow data were collected from the existing telemetry sys- costs are estimated to be 13.18 TIRR/SC/year (0.314 USD/SC/
tem, which was logged at the three inlet points to zone J year). It should be mentioned that the collection and analysis
every 15 min for 24 h. Data cleaning, validation, and nor- of cost data were conducted during the research period
malization were performed using Python to achieve a reli- between Mar. 2017 and Mar. 2018, at that time, approximately
able data set. Figure 5 reports the average daily water inflow USD 1 ¼ 42 TIRR.
for zone J during the period between Mar. 2017 and Mar. The marginal cost of ALC can vary depending on how
2018. The average daily inflow of the zone varies between much effort is being carried out by the water utility. In this
393.20 and 1122 m3/hour (average value of 747.4). As Figure study, the marginal cost of ALC, using Equation (8), is
5 shows, zone J characterized by remarkable seasonal found to be 1.37 TIRR/m3 (0.033 USD/m3) under current
changes in water consumption because the network serves operating conditions. The SR-MCW, including the cost of
mainly residential users. The pressure step test was under- power and chemicals, is estimated from the company’s exist-
taken on September 21, 2017, to estimate the FAVAD N1 ing water sources. Considering these two components, the
exponent. The pressure was reduced in three steps during estimated MCW for MWWC is found to be 6 TIRR/m3
the 2:00 to 5:00 a.m. The reduction of flow into the zone (0.143 USD/m3). One conclusion derived from the estimated
and pressure changes at the AZP were recorded. The aver- marginal costs is that the water utility can spend an add-
age value of N1 was estimated to be 1.14. itional annual expenditure of 4.63 TIRR/m3 (0.11 USD/m3)
STRUCTURE AND INFRASTRUCTURE ENGINEERING 829

Table 1. General characteristics of WDN of zone J.


Length By Pipe Material Length By Pipe Diameter Length By Pipe Age
Material Length(km) % Diameter(mm) Length(Km) % Age (years) Length(Km) %
Sub Zone J1Long (Km)¼59.9, Area (Km2)¼7.70 5L 0.013 0.02 <100 30.191 50.34 12 0.380 0.63
ACC 28.184 46.99 100-150 17.012 28.36 11 0.000 0.00
CI 0.006 0.01 200-250 3.000 5.00 10 0.264 0.44
DCI 14.284 23.81 300-350 0.552 0.92 9 38.703 64.53
GRP 0.776 1.29 400-450 8.159 13.60 8 1.334 2.22
NP 0.018 0.03 500-550 0.000 0.00 7 0.204 0.34
PE 16.094 26.83 600-650 0.420 0.70 6 2.200 3.67
PP 0.028 0.05 700-750 0.477 0.80 5 2.560 4.27
PVC 0.152 0.25 >750 0.085 0.14 4 0.429 0.72
ST 0.424 0.71 unknown 0.069 0.12 3 0.413 0.69
unknown 0.009 0.01 2 0.018 0.03
unknown 13.481 22.48
Sub Zone J2Long (Km)¼52.6, Area (Km2)¼4.54 AC 0.115 0.22 <100 11.403 21.68 33 0.042 0.08
ACC 37.139 70.60 100-150 27.652 52.57 18 0.268 0.51
DCI 5.521 10.50 200-250 8.074 15.35 11 0.279 0.53
PE 9.471 18.00 300-350 0.025 0.05 10 0.694 1.32
ST 0.252 0.48 400-450 2.773 5.27 9 45.432 86.37
unknown 0.104 0.20 500-550 0.892 1.70 8 0.032 0.06
>600 1.620 3.08 7 0.054 0.10
UNK 0.058 0.11 6 0.716 1.36
unknown 0.104 0.20 5 0.030 0.06
4 0.074 0.14
unknown 4.979 9.47

Figure 5. Average daily water inflow for zone J during the period between Mar. 2017 and Mar. 2018.

Table 2. ALC costs data obtained from the MWWC.


Cost Symbol Value
Unit cost of Leakage survey cost for mains (TIRR/ km length of mains) CLM 4000
Length of Mains (km) LM 112.5
Unit cost of Leakage survey for service connections (TIRR/ connections) CNC 20
Number of service connections NC 26566
Unit cost of the bonus payment to the contractor for detected leaks on mains (TIRR/detected leaks) CBp  main 2300
Unit cost of repaired leaks on mains (TIRR/repaired leaks) CRp  main 8800
Unit cost of extra payment to the contractor for detected leaks on service connections (TIRR/ detected leaks) CBp  conn: 1800
Unit cost of repaired leaks on service connections (TIRR/ repaired leaks) CRp  conn: 2500
Number of detected and repaired leaks on mains NumM 20
Number of detected and repaired leaks on service connections NumC 44
Total variable Costs (TIRR/ SC/ year) 52.45

for reducing leakage levels to SR-ELL under current operat- losses. Flow monitoring reveals that the minimum value of
ing conditions. night flow occurs in the time interval between 02:00 a.m.
and 5:00 a.m. Furthermore, as customer meters are located
4.2. Minimum night flow analysis inside the study areas properties, no minimum night con-
sumption readings could be taken. Therefore, the CNU is
In this study, based on 24-hour zone measurements, an estimated based on the customer night consumption for
MNF analysis is performed to estimate the volume of real other countries, as reported in previous studies (Amoatey
830 I. MOSLEHI ET AL.

Figure 6. MNF and total inflow variations for zone J during the research period.

Table 3. Results of MNF analysis and estimation of the current level of leakage for zone J during the research period.
NNF(m3/h) NDF Daily Leakage Level (m3/day) Annual Leakage Level (MCM/year) Annual Inflow Rate (MCM/year) Average Leakage (%)
167.1 21.43 3581 1.31 6.49 20.2

Table 4. Calculations of maximum and economic potentially recover- 4.3. Estimation of CARL, UARL and UBL
able leakage.
Parameter Symbol Units Value % of Total The calculations of UARL and UBL for zone J are presented
Current Annual Real Losses CARL m3/SC/year 49.2 – in Table 4. As can be seen, the calculated value of UARL
Short-run Economic SR-ELL m3/SC/year 27 55.9
Leakage Level
using Equation (4) is found to be 0.377 MCM/year, which
Target Background TBL m3/SC/year 20.35 41.4 corresponds to 14.2 m3/SC/year. The calculated value of ILI
Leakage, if ICF ¼ 2 for the WDN of zone J is then obtained to be 3.46. With
Maximum potentially MPRL m3/SC/year 28.85 58.6
Recoverable Leakage
this ILI value, the network is classified as A2 based on real
Economic potentially EPRL m3/SC/year 22.2 44.1 losses assessment matrix (WBI target matrix) for low and
Recoverable Leakage middle-income countries (Liemberger et al., 2007;
Liemberger, 2010). Consequently, the improvement of ALC
activities and assessment of ELL is required.
et al., 2014; Morrison, Tooms, & Rogers, 2007). The cus- The TBL can be assessed by multiplying the calculated UBL
tomer night use is assumed to be equal to 5 lit/SC/hour, by an estimation of ICF. The value of UBL using Equation (5)
similar to reported CNU for Malaysia (Amoatey et al., 2014; is calculated as 0.27 MCM/year or 10.17 m3/SC/year (Table 4).
Fantozzi & Lambert, 2012). This value is chosen because As mentioned previously, there are several methods for estimat-
Malaysia is an Islamic country with similar social and reli- ing ICF. In this paper, ICF is obtained based on the estimation
gious costumes. Moreover, there are no large night users of target ILI. An initial estimate for ICF is considered approxi-
(industrial and agricultural) in the studied area. mately equal to the ILI value when the target level of leakage
Figure 6 plots variation in MNF and total inflows into the has been achieved. In addition, the target ILI can be selected
studied DMA during the research period. The average MNF dur- using the current ILI value as well as preliminary leakage man-
ing this period is calculated to be 299.9 m3/hourwhich is approxi- agement target-setting guidelines (AWWA, 2016). In this guide-
mately 40% of the average flow rate into zone J. This analysis line, the target ILI range is identified based on the water
indicates that the share of leakage may be significant in zone J. resources’ circumstances, operational and financial considera-
Moreover, the average level of NNF is calculated to be 167.1 m3/ tions that water utility encountered.
hour. It should be noted that the calculated NNF is mostly com- Accordingly, it is found that the target ILI range of 1.0-
posed of real losses due to leakage in the water mains and service 3.0 most closely represents the conditions in the water util-
connections, as flow rates are measured at the inlet points of the ity based on the calculated ILI value (ILI ¼ 3:46) and also
network. The value of the NDF, using Equation (3), is calculated the circumstances encountered by the MWWC under the
to be 21.43. Therefore, the estimated daily leakage derived by three aforementioned considerations. Therefore, for the ini-
MNF analysis is found to be 3582 m3 per day, with correspond- tial estimation of ICF, the mid-level of the determined target
ing annual leakage of 1.31 MCM/year. In terms of service con- ILI range (i.e., ILITarget ¼ 2) can be used. The resulting TBL
nections, the current annual leakage is equal to 49.2 m3/SC/year. for zone J is then obtained to be 0.54 MCM/year or 20.35
Table 3 reports the summary of night flow analysis for zone J. As m3/SC/year. Eventually, the PLL can be estimated using
can be seen in this table, the leakage accounts for approximately NNF data during the same period of MNF analysis by cal-
20.2% of the total annual inlet volume. culating the maximum of the 7-day rolling 50th percentile
STRUCTURE AND INFRASTRUCTURE ENGINEERING 831

Figure 7. Analysis of ELL for zone J under current operating conditions.

values (UKWIR, 2011). Following this procedure, the PLL for zone J is found to be 22.2 m3/SC/year, which is equiva-
for zone J is estimated to be 129.3 m3/SC/year. lent to 0.59 MCM/year (Table 4). Looking at Table 4, the
EPRL comprises 45% of the CARL. This means that there is
an apparent justification for the further reduction in the
4.4. Economic level of leakage estimation
leakage level in zone J under the current ALC policy and
Figure 7 represents the steady-state ALC cost curve (solid operating pressure. From an economic point of view, if the
line), the cost curve of water loss (dashed line) at different CARL is reduced by 45% (i.e., reaches 11% of supplied
leakage levels, and the total cost curve (point line) for zone water), the ALC policy can be quite effective.
J under current operating conditions. As may be observed As shown in Figure 7, the annual economic budget for
in Figure 7, the ALC cost curve tends towards infinity at ALC activities is obtained to be 285 TIRR/SC/year (6.786
the TBL of 20.35 m3/SC/year. The maximum potentially USD/SC/year), that is, around four times higher than the
recoverable leakage (MPRL) can be obtained by deducting current ALC costs which are approximately 65.63 TIRR/SC/
the estimated TBL from the CARL. The MPRL is found to year (1.563 USD/SC/year). These results suggest that the
be 28.85 m3/SC/year, which can be achieved with unlimited intervention frequency should be performed economically
ALC resources for the study area. The curve also falls to the every 3 months in the study area to achieve SR-ELL. In add-
minimum level of 13.18 TIRR/SC/year (0.314 USD/SC/year), ition, the results indicate that the value of water lost in this
which is equivalent to the leakage monitoring costs (ALC case is high enough to justify more frequent interventions
fixed costs) at the passive leakage level of 129 m3/SC/year. for further leakage reductions. This is due to the fact that
As can be seen in Figure 7, the value of SR-ELL for zone J the average run time of unreported leakages can be reduced
is obtained as 27 m3/SC/year that corresponds to 0.717 by increasing the annual operating budget for ALC activities
MCM/year at the current operating pressure and current and thereby, economic intervention frequencies.
ALC policy adopted by the MWWC.
The estimated SR-ELL can be set as the target level of
4.5. Sensitivity analysis
leakage, which can be achieved by the water utility under
current operating conditions. In order to control the ICF In the last step of the proposed methodology, the sensitivity
estimation method, the economic ILI is calculated by divid- analysis is carried out to examine the variables affecting SR-
ing the estimated SR-ELL by the calculated UARL and com- ELL. The representative variables applied to this study are
pared with the determined target ILI, obtained from leakage average operating pressure, marginal cost of water, infra-
management target-setting guidelines. Following this pro- structure condition, and the marginal cost of ALC activity.
cedure, the economic ILI (new target ILI) is 1.94, which is The necessity of incorporating PM options into the estima-
reasonably close to the determined target ILI of 2.0. Thus, it tion of SR-ELL has been widely accepted (Fanner &
may be concluded that the estimated ICF is acceptable and Lambert, 2009; Fantozzi & Lambert, 2007). Pressure varia-
it is not necessary to update the estimated ICF. tions can affect the leak flow rates of all components of
The economic potentially recoverable leakage (EPRL) is leakage, pipe break frequencies, and thereby SR-ELL. After
obtained by deducting CARL from SR-ELL. The EPRL value implementing PM, the BL, the frequency, and the flow rate
832 I. MOSLEHI ET AL.

Figure 8. The relationship between ALC costs and different operating pressures for five specific leakage levels.

of reported and unreported leakages are reduced. In con- The estimated pipe breaks frequency (mains and service
trast, if operating pressure is increased to improve service, connections) and the predicted leak flow rates (i.e., back-
new leaks may occur, and flow rates from existing leaks ground, passive and current level of leakage) due to imple-
can increase. menting PM are then used to calculate ALC costs. Figure 8
As mentioned previously, the steady-state ALC cost curve shows the relation between ALC costs and different operat-
is developed for modelling steady-state ALC costs for leak- ing pressures for five specific leakage levels.
age levels above and below the current level under current As Figure 8 shows, all ALC costs for different leakage
operating pressure and ALC policy. Therefore, the ALC cost levels follow a common trend: the ALC costs are higher for
curves for different operating pressures can be developed high values of operating pressure values and decrease for
using pressure-leakage and pressure-pipe breaks relation- small values. This means that lower ALC costs are required
ships. Changes in leak flow rates (L0 to L1 ) as average oper- to maintain leakage at a specific steady-state level after
ating pressure at the AZP varies (AZP0 to AZP1 ) is implementing PM schemes. For example, at the current level
estimated using N1 power law, an approximate version of of leakage, 49.2 m3/SC/year, if the average operating pressure
the FAVAD concept, with the following equation: L0 =L1 ¼ can be reduced by 44% (from 44.5 m to 25 m), the predicted
ðAZP0 =AZP1 ÞN1 (Lambert & Fantozzi, 2010; Thornton & reduction of ALC costs will be around 65% (66 to 23 TIRR/
Lambert, 2005). It is important to note that in this equation, SC/year (1.571 to 0.548 USD/SC/year)). Moreover, the rate
the value of N1 exponent for estimating BL changes should of increase in ALC costs is not constant, and it is much
be assumed to be close to 1.5, as the BL is very sensitive to greater when the average operating pressure is higher. This
pressure (Fantozzi & Lambert, 2007). may be expected given that a change in pressure can
Furthermore, the relationship between pipe breaks fre- increase the pipe break frequencies with a cubic relation-
quency and maximum pressure at the AZP can be predicted ship; variable costs of ALC will increase nonlinearly.
using the following equation (Lambert, Fantozzi, & Figure 9 represents the influence of pressure changes for
Thornton, 2013; Lambert & Thornton, 2012; Moslehi and different SR-MCW on the SR-ELL. As can be seen, the SR-
Jalili_ghazizadeh, 2020): ELL shows a decreasing trend as the operating pressure
  ! reduce. This may be owing to the fact that implementing
AZPmax1 N2 PM can cause lower pipe break rates, lower leak flow rates
ðBF0  BF1 Þ ¼ ðBF0  BFnpd Þ  1 
AZPmax2 specifically for BL, lower NRR of unreported leakage, and
thereby lower SR-ELL (Fantozzi & Lambert, 2007;
(9)
Ghorbanian et al., 2016). It is also relevant to note that the
where, BF0 and BF1 ¼ pipe breaks frequency before and SR-ELL rate is not directly proportional to pressure, and it
after PMrespectively, BFnpd ¼ pressure-independent pipe is lower when the operating pressure reduces. For the
break frequency; AZPmax1 and AZPmax2 ¼ maximum pres- MWWCʼs current operating condition, a 30% reduction in
sure at the AZP before and after pressure changes, respect- SR-ELL requires a 20% reduction in the operating pressure.
ively. N2 ¼ the exponent of pressure-pipe breaks Figure 9 also demonstrates that for different operating pres-
relationship is recommended to be close to 3 (Ghorbanian, sure, the SR-ELL is not sensitive to the SR-MCW up to a
Guo, & Karney, 2016; Lambert & Thornton, 2012). certain value.
STRUCTURE AND INFRASTRUCTURE ENGINEERING 833

Figure 9. The influence of pressure variations for different SR-MCW on the SR-ELL.

Figure 10. The relation between the total costs (ALC costs plus the cost of water lost) and leakage levels for different SR-MCW.

The influences of the SR-MCW on the SR-ELL are also m3), the SR-ELL grows by approximately 30%. This can be
assessed at the current operating pressure and ALC activity. linked to the fact that the ALC activity follows the law of
Figure 10 presents the relation of the total annual costs diminishing returns which results in a slowing down in the
(ALC costs plus the cost of water lost) versus leakage levels rate of reduction of SR-ELL when more resources (higher
for different SR-MCW. The figure indicates that the SR- value of SR-MCW) are dedicated to leakage reduction
MCW has little influence on the SR-ELL up to a certain (Pearson & Trow, 2005; Thornton et al., 2008).
value. As can be seen, if the SR-MCW is more than 6 Another key variable in the estimation of SR-ELL is the
TIRR/m3 (0.143 USD/m3), the SR-ELL does not change sig- condition of infrastructure, which directly influences the
nificantly, while the total costs of ALC will increase signifi- volume of BL (AWWA, 2016; Fanner et al., 2007). Figure
cantly with increasing SR-MCW. For example, at the 11 relates the SR-ELL for different values of TBL and SR-
current operating pressure (i.e., an average pressure of MCW under current ALC policy and operating pressure.
44.5 m) and ALC activity, doubling the SR-MCW from 6 to TBL levels are estimated by multiplying the calculated UBL
12 TIRR/m3 (0.143 to 0.286 USD/m3) only offers a 10% by different ICF values, varying from infrastructure good
reduction in SR-ELL. However, when the SR-MCW is condition (ICF ¼ 1) to poor condition (ICF ¼ 3). Figure 11
reduced by 50% from 4 to 2 TIRR/m3 (0.095 to 0.048 USD/ reveals a direct link between SR-ELL and TBL. When
834 I. MOSLEHI ET AL.

Figure 11. The influence of infrastructure condition on SR-ELL for different SR-MCW.

Figure 12. The influence of variable elements of ALC costs on SR-ELL for different SR-MCW at the current operating pressure.

infrastructure condition gradually worsens, the ICF and The relationship between SR-ELL, variable costs of ALC
thereby, the minimum achievable leakage level (i.e., TBL), activity, and SR-MCW is represented in Figure 12. The
which can be achieved through limitless ALC costs, will influence of variable elements of ALC costs, including detec-
increase. The higher level of TBL can cause moving the tion and repair costs, are assessed for different SR-MCW at
ALC cost curve and also total cost curve to the right the current operating pressure. Figure 12 shows a common
because of the increased collective sum of minor leaks that trend for different SR-MCW: the SR-ELL is lower for small
are too small to be found by the current ALC policy. values of variable costs of ALC and would increase for high
Therefore, the SR-ELL for the current operating conditions values. This is due to the fact that for a specific SR-MCW,
can increase when the condition of the network infrastruc- the marginal cost of ALC increase for the higher values of
ture gradually worsens. variable ALC costs, and therefore the SR-ELL increases for a
Moreover, as may be observed in Figure 11, the SR-ELL certain value of water lost. However, the rate of increase of
is almost linearly proportional to the TBL. For example, If SR-ELL is not constant for different SR-MCWs, and it is
the infrastructure condition of the network changes from reduced for the higher SR-MCW. For example, for a higher
current average conditions (ICF ¼ 2) to poor conditions SR-MCW value, 12 TIRR/m3 (0.286 USD/m3), the SR-ELL
(ICF ¼ 3), then the SR-ELL will increase approximately by remains largely indifferent to the ALC variable costs,
33%. Moreover, the volume of BL comprises a high propor- whereas, for a lower value of the SR-MCW, 2 TIRR/m3
tion of the current leakage level (more than 40%) for zone (0.048 USD/m3), the SR-ELL is considerably affected by
J. Thus, SR-ELL may be significantly reduced by improving ALC variable costs.
infrastructure conditions through leakage control activities
like network rehabilitation, renewal and replacement (i.e.,
5. Conclusions
asset management). In contrast, if the infrastructure condi-
tion will improve to good conditions (ICF ¼ 1), then the This paper presents a data-based field methodology that
SR-ELL reduces from 27 to 17.8 m3/SC/year for the current aims at estimating the short-run economic level of leakage
operating conditions. concerning active leakage control, which can be utilized by
STRUCTURE AND INFRASTRUCTURE ENGINEERING 835

water utilities to have a clear strategy on their leakage con- significantly influences the short-run economic level of leak-
trol policies. The methodology was applied to a large indi- age, as it directly affects the background leakage, which
vidual zone Mashhad (Iran) that operated under steady-state comprises of more than 40% of the current leakage in the
conditions. The average zone points were determined using study area. The proposed methodology could assist water
a systematic approach and GIS tools for assessing the aver- utilities in deciding the leakage target level based on the
age operating pressure of the case study. economic level of leakage. It offers a practical approach in
Different leakage levels were estimated using collected estimating short-run economic leakage level with respect to
flow and pressure data to develop an active leakage control the active leakage control strategy.
cost curve. Night flow analysis was carried out to estimate
the night leakage rate. Subsequently, the night-day factor
was calculated to predict average daily leakage from night Acknowledgments
leakage rate. Night flow results reveal that the current leak-
age level is 49.2 m3/SC/year, which is around 20% of the The authors appreciate the help provided by the Mashhad water and
wastewater company through provision of data used in this work and
total input volume of water. The target background leakage,
scientific and technological support.
which is achievable with the current policies and unlimited
active leakage control resources, is estimated to be 20.35 m3/
SC/year. The passive leakage level is found to be 129.3 m3/
SC/year when the water utility performs no active leakage Disclosure statement
control activity. No potential conflict of interest was reported by the author(s).
After developing the active leakage control cost curve
and total cost curve, the short-run economic level of leakage Notations
is estimated to be 27 m3/SC/year for the studied area at the ALC Active Leakage Control
current operating pressure and active leakage control policy. AZNP Average Zone Night Pressure
AZP Average Zone Point
The economic intervention frequency of active leakage con- BABE Burst and Background Estimates
trol should be undertaken every three months to achieve the BL Background Leakage
estimated short-run economic level of leakage. Results show CARL Current Annual Real Losses
that when the current leakage level is reduced to short-run DEM Digital Elevation Model
DMA District Metered Area
economic level of leakage through active leakage control EIF Economic Intervention Frequency
activity, 45% of the total leakage can be potentially recover- ELL Economic Level of Leakage
able, and the ILI will be proportionally decreased from 3.46 EPRL Economic Potentially Recoverable Leakage
to 1.94. From an economic point of view, an additional EURL Economic Unavoidable Real Losses
annual expenditure of 4.63 TIRR (USD 0.086) per cubic FAVAD Fixed and Variable Area Discharge
GIS Geographic Information System
metre of water can be spent by the water utility for reducing ICF Infrastructure Condition Factor
leakage levels under current operating conditions. ILI Infrastructure Leakage Index
The effects of different influential factors on the short- ILIMSS International Leakage Management Support Services
run economic level of leakage and their interrelationships IWA International Water Association
LNC Legitimate Night-time Consumption
were investigated by conducting sensitivity analysis. The
LR-ELL Long-run Economic Level of Leakage
sensitivity analysis results prove that implementing pressure LRMC Long-run Marginal Cost
management has a significant influence on the short-run MNF Minimum Night Flow
economic level of leakage, as the average operating pressure MPRL Maximum Potentially Recoverable Leakage
in the study area was relatively high. Therefore, the imple- MWWC Mashhad Water and Wastewater Company
NDF Night-Day Factor
mentation of pressure management can cause less frequent NNF Net Night Flow
pipe breaks and reduce leakage components, especially back- NPV Net Present Value
ground leakage. For the current operation, if the operating NRR Natural Rate of Rise
pressure is reduced by 20%, the short-run economic level of PLC Passive Leakage Control
leakage will decrease by 30%. PLL Passive Level of Leakage
PM Pressure Management
Results demonstrate that the short-run economic level of PRV Pressure Reducing Valves
leakage remains unaffected for the short-run marginal cost RR Rate of Rise
of water up to a certain level. This may be due to the fact SC Service Connections
that the active leakage control activity follows the law of SCADA Supervisory Control and Data Acquisition
SELL Sustainable Economic Level of Leakage
diminishing returns, which means that, as more money is
SR-ELL Short-run Economic Level of Leakage
spent on active leakage control activity, units of water saved SRMC Short-run Marginal Cost
for each unit of money spent will be less. The short-run SR-MCW Short-run Marginal Cost of Water
economic level of leakage can only reduce with 10% by dou- TBL Target Background Leakage
bling the short-run marginal cost of water (from 6 to 12 TIRR Thousands of IR Rials
UARL Unavoidable Annual Real Losses
TIRR/m3 (0.143 to 0.286 USD/m3)) at the current operating UBL Unavoidable Background Leakage
pressure and active leakage control policy. The results also WAGL Weighted Average Ground Level
point to this fact that the infrastructure condition WDN Water Distribution Network
836 I. MOSLEHI ET AL.

References expenditures and revenue losses. In IWA Balkan Young Water


Professionals Conference (pp. 182–190). Thessaloniki, Greece.
Al-Washali, T., Sharma, S., & Kennedy, M. (2016). Methods of assess- Gonelas, K., & Kanakoudis, V. (2016). Reaching economic leakage level
ment of water losses in water supply systems: a review. Water through pressure management. Water Supply, 16(3), 756–765.
Resources Management, 30(14), 4985–5001. https://doi.org/10.1007/ https://doi.org/10.2166/ws.2015.181
s11269-016-1503-7 Gupta, A., & Kulat, K. D. (2018). A selective literature review on leak
Alkasseh, J. M. A., Adlan, M. N., Abustan, I., Aziz, H. A., & Hanif, management techniques for water distribution system. Water
A. B. M. (2013). Applying minimum night flow to estimate water Resources Management, 32(10), 3247–3269. https://doi.org/10.1007/
loss using statistical modeling: A case study in Kinta Valley. Water s11269-018-1985-6
Resources Management, 27(5), 1439–1455. https://doi.org/10.1007/ Haider, H., Al-Salamah, I. S., Ghazaw, Y. M., Abdel-Maguid, R. H.,
s11269-012-0247-2 Shafiquzzaman, M., & Ghumman, A. R. (2019). Framework to
Alkasseh, J. M. A., Adlan, M. N., Abustan, I., & Hanif, A. B. M. Establish Economic Level of Leakage for Intermittent Water
(2015). Achieving an economic leakage level in Kinta Valley, Supplies in Arid Environments. Journal of Water Resources Planning
Malaysia. Water Utility Journal, 11, 31–47. and Management, 145(2), 05018018. https://doi.org/10.1061/
Amoatey, P. K., Minke, R., & Steinmetz, H. (2014). Leakage estimation (ASCE)WR.1943-5452.0001027.
in water networks based on two categories of night-time users: A Hamilton, S., & Mckenzie, R. (2014). Water management and water
case study of a developing country network. Water Supply, 14(2), loss. Water Intelligence Online (Vol. 13). London: IWA Publishing.
329–336. https://doi.org/10.2166/ws.2013.201 https://doi.org/10.2166/9781780406367
Amoatey, P. K., Minke, R., & Steinmetz, H. (2018). Leakage estimation Howarth, D. A. (1998). Arriving at the economic level of leakage:
in developing country water networks based on water balance, min- Environmental aspects. Water and Environment Journal, 12(3),
imum night flow and component analysis methods. Water Practice 197–201. https://doi.org/10.1111/j.1747-6593.1998.tb00173.x
and Technology, 13(1), 96–105. https://doi.org/10.2166/wpt.2018.005 ILMSS Ltd. (2013). Guidelines relating to the assesment and calculation
Ashton, C. H., & Hope, V. S. (2001). Environmental valuation and the of average pressure in water distribiution systems and zones.
economic level of leakage. Urban Water, 3(4), 261–270. https://doi. Gwynedd, UK: ILMSS.
org/10.1016/S1462-0758(01)00046-2 Islam, M. S., & Babel, M. S. (2013). Economic analysis of leakage in
American Water Works Association. (2016). Water Audits and Loss the Bangkok Water distribution system. Journal of Water Resources
Control Programs: M36. Denver: American Water Works Planning and Management, 139(2), 209–216. https://doi.org/10.1061/
Association. (ASCE)WR.1943-5452.0000235
Bettin, A., Rogers, D., & Serrani, C. (2014). Beyond leakage manage- Kanakoudis, V., & Gonelas, K. (2015). The optimal balance point
ment: How to decide if, where and when to look for leaks. Water between NRW reduction measures, full water costing and water
Utility Journal, 7, 45–49. pricing in water distribution systems. Alternative scenarios forecast-
Chisakuta, S., Mayumbelo, K., Mulenga, K., Simbeye, I., Wegelin, W., ing the Kozani’s WDS optimal balance point. Procedia engineering,
McKenzie, R., & Andres, D. (2011). Non-revenue water: trainers 119(2015), 1278–1287. https://doi.org/10.1016/j.proeng.2015.08.996
manual. Wave Pool Zambia Imprint. Germany: German Foundation Kanakoudis, V., & Gonelas, K. (2016). Analysis and calculation of the
for International Cooperation (GIZ). short and long run economic leakage level in a water distribution
Creaco, E., & Walski, T. (2017). Economic analysis of pressure control system. Water Utility, 12, 57–66.
for leakage and pipe burst reduction. Journal of Water Resources Kanakoudis, V., & Muhammetoglu, H. (2014). Urban water pipe net-
Planning and Management, 143(12), 04017074. https://doi.org/10. works management towards non-revenue water reduction: Two case
1061/(ASCE)WR.1943-5452.0000846 studies from Greece and Turkey. CLEAN - Soil, Air, Water, 42(7),
Fanner, P., & Lambert, A. (2009, April). Calculating SRELL with pres- 880–892. https://doi.org/10.1002/clen.201300138
sure management, active leakage control and leak run-time options, Kanakoudis, V., & Tsitsifli, S. (2010). Results of an urban water distri-
with confidence limits. In Proceedings of 5th IWA Water Loss bution network performance evaluation attempt in Greece. Urban
Reduction Specialist Conference (pp. 373–380). Water Journal, 7(5), 267–285. https://doi.org/10.1080/1573062X.
Fanner, P. V., Strum, R., Thornton, J., Liemberger, R., Davis, S. E., & 2010.509436
Hoogerwerf, T. (2007). Leakage management technologies. American Kanakoudis, V., Tsitsifli, S., Cerk, M., Banovec, P., Samaras, P., &
Water Works Association. Denver, Colo: Awwa Research Zouboulis, A. I. (2015). Basic principles of a DSS tool developed to
Foundation. prioritize NRW reduction measures in water pipe networks. Water
Fantozzi, M., & Lambert, A. (2007). Including the effects of pressure Quality, Exposure and Health, 7(1), 39–51. https://doi.org/10.1007/
management in calculations of Short-Run Economic Leakage Levels. s12403-014-0111-0
In IWA Conference’Water Loss (pp. 1–10). Bucharest, Romania. Kanakoudis, V., Tsitsifli, S., & Demetriou, G. (2016). Applying an inte-
Fantozzi, M., & Lambert, A. (2012). Residential night consumption – grated methodology toward non-revenue water reduction: the case
Assessment, choice of scaling units and calculation of variability. In of Nicosia, Cyprus. Desalination and Water Treatment, 57(25),
IWA Conference “Water Loss 2012 (pp. 1–10). Manila, Philippines. 11447–11461. https://doi.org/10.1080/19443994.2015.1048537
Farley, M., & Liemberger, R. (2005). Developing a non-revenue water Kanakoudis, V., Tsitsifli, S., Kouziakis, C., & Lappos, S. (2015).
reduction strategy: planning and implementing the strategy. Water Defining the level of the non-revenue water in Kozani, Greece: is it
Supply, 5(1), 41–50. https://doi.org/10.2166/ws.2005.0006 a typical case? Desalination and Water Treatment, 54(8), 2170–2180.
Farley, M., & Trow, S. W. (2007). Losses in water distribiution net- https://doi.org/10.1080/19443994.2014.934119
works; a practitioners guide to assessment, monitoring, and control Kanakoudis, V., Tsitsifli, S., Samaras, P., & Zouboulis, A. (2013).
(2nd ed.). UK: International Water Association Publishing. Assessing the performance of urban water networks across the EU
Frauendorfer, R., & Liemberger, R. (2010). The issues and challenges of Mediterranean area: The paradox of high NRW levels and absence
reducing non-revenue water. Metro Manila, Philippines: Asian of respective reduction measures. Water Supply, 13(4), 939–950.
Development Bank. https://doi.org/10.2166/ws.2013.044
Ghorbanian, V., Guo, Y., & Karney, B. (2016). Field data–based meth- Kanakoudis, V., Tsitsifli, S., & Zouboulis, A. I. (2015). WATERLOSS
odology for estimating the expected pipe break rates of water distri- project: developing from theory to practice an integrated approach
bution systems. Journal of Water Resources Planning and towards NRW reduction in urban water systems. Desalination and
Management, 142(10), 04016040. https://doi.org/10.1061/(ASCE)WR. Water Treatment, 54(8), 2147–2157. https://doi.org/10.1080/
1943-5452.0000686 19443994.2014.934114
Gonelas, K., & Kanakoudis, V. (2015). The economic impact of pressure Lambert, A., & McKenzie, R. D. (2002). Practical experience in using
management in Kozani city’s water distribution system. Benefits, the infrastructure leakage index. In IWA Conference “Leakage
STRUCTURE AND INFRASTRUCTURE ENGINEERING 837

Management: A Practical Approach (Vol. 3, pp. 1). Lemesos, and anomalous event detection: A practical approach. Urban Water
Cyprus. Journal, 13(3), 242–252. doi:10.1080/1573062X.2014.988733.
Lambert, A. (2001). What do we know about pressure-leakage relation- Mazzolani, G., Berardi, L., Laucelli, D., Simone, A., Martino, R., &
ships in distribution systems. In IWA Conference. Systems approach Giustolisi, O. (2017). Estimating leakages in water distribution net-
to leakage control and water distribution system management (pp. works based only on inlet flow data. Journal of Water Resources
1–8). Planning and Management, 143(6), 04017014–04017011. https://doi.
Lambert, A. (2002). International report: Water losses management org/10.1061/(ASCE)WR.1943-5452.0000758
and techniques. Water Supply, 2(4), 1–20. https://doi.org/10.1177/ Morrison, J., Tooms, S., & Rogers, D. (2007). District metered areas
001789695501300409 guidance notes. London: IWA Publishing.
Lambert, A. (2009). Ten years experience in using the UARL formula to Moslehi, I., & Jalili_ghazizadeh, M. (2020). Pressure-Pipe Breaks
calculate infrastructure leakage index. In IWA Conference “Water Relationship in Water Distribution Networks: A Statistical Analysis.
Loss 2009 (pp. 189–196). Water Resources Management, 1–18. 10.1007/s11269-020-02587-4
Lambert, A., Charalambous, B., Fantozzi, M., Kovac, J., Rizzo, A., St Moslehi, I. (2020). Developing a decision support system for real loss
John, S. G., & Galea St. John, S. (2014). 14 years experience of using management in water distribution networks (Doctoral dissertation).
IWA best practice water balance and water loss performance indica- Department of Civil, Water, and Environmental Engineering,
tors in Europe. In IWA Specialized Conference: Water Loss (pp. Shahid Beheshti University, Tehran, Iran (In Persian).
1–31). Munoz-Trochez, C., Smout, I., & Kayaga, S. (2011). Economic level of
Lambert, A., & Fantozzi, M. (2005). Recent advances in calculating leakage (ELL) calculation with limited data: an application in
economic intervention frequency for active leakage control, and Zaragoza. In 35th WEDC International Conference. Loughborough,
implications for calculation of economic leakage levels. Water UK.
Supply, 5(6), 263–271. https://doi.org/10.2166/ws.2005.0072 Mutikanga, H. E., Sharma, S. K., & Vairavamoorthy, K. (2013).
Lambert, A., & Fantozzi, M. (2010). Recent developments in pressure Methods and tools for managing losses in water distribution sys-
management. In IWA Conference ‘Water Loss. tems. Journal of Water Resources Planning and Management, 139(2),
Lambert, A., Fantozzi, M., & Shepherd, M. (2017). CCWi2017: F100 166–174. https://doi.org/10.1061/(ASCE)WR.1943-5452.0000245
‘Pressure: Leak Flow Rates Using FAVAD: An Improved Fast-Track National water and Wastewater Engineering Company. (2018, July 20).
Practitioner’s Approach’ (Version 1). The University of Sheffield. Annually non-revenue water reports on water distribution system.
https://doi.org/10.15131/shef.data.5365078.v1 Retrieved from https://www.nww.ir/
Lambert, A., Fantozzi, M., & Thornton, J. (2013). Practical approaches Pearson, D., & Trow, S. W. (2005). Calculating the economic levels of
to modeling leakage and pressure management in distribution system- leakage. In Leakage Conference (pp. 1–16). Portugal.
s–progress since 2005. In 12th International Conference on Puust, R., Kapelan, Z., Savic, D. A., & Koppel, T. (2010). A review of
Computing and Control for the Water Industry (pp. 2–4). Perugia, methods for leakage management in pipe networks. Urban Water
Italy. Journal, 7(1), 25–45. https://doi.org/10.1080/15730621003610878
Lambert, A., Koelbl, J. K., & Fuchs-Hanusch, D. (2014). Interpreting Renaud, E., Sissoko, M. T., Clauzier, M., Gilbert, D., Sandraz, A. C., &
ILIs in small systems. In IWA Water Ideas 2014: Intelligent Pillot, J. (2015). Comparative study of different methods to average
Distribution for Efficient and affordable Supplies (pp. 1–10). pressures in water distribution zones. Water Utility Journal, 10,
Lambert, A., & Lalonde, A. (2005). Using practical predictions of eco- 25–35.
nomic intervention frequency to calculate short-run economic leakage Smout, I. K., Kayaga, S., & Munoz-Trochez, C. (2010). Adapting the
level, with or without pressure management. In IWA Specialised economic level of leakage concept to include carbon emissions, and
Conference ‘Leakage (pp. 310–321). application with limited data. In IWA World Water Congress 2010
Lambert, A., & McKenzie, R. D. (2002). Practical experience in using (pp. 1–10).
the infrastructure leakage index. In IWA Conference “Leakage Thornton, J., & Lambert, A. (2005). Progress in practical prediction of
Management: A Practical Approach (Vol. 3, pp. 1–16). Lemesos, pressure: leakage, pressure: burst frequency and pressure: consumption
Cyprus. relationships. In IWA Special Conference’Leakage (pp. 1–10).
Lambert, A., Merks, C., & Trow, S. (2015). EU reference document Thornton, J., Sturm, R., & Kunkel, G. (2008). Water loss control.
good practices on leakage management WFD CIS WG PoM: Main McGraw Hill Professional. New York: McGraw-Hill.
Report. Publications Office. https://doi.org/10.2779/102151 Tripartite Group (2002). Best practice principles in the economic level of
Lambert, A., & Taylor, R. (2010). Water loss guide, a guideline for the leakage calculation. UK: Tripartite Group.
New Zealand water and wastes association. New Zealand: Water Trow, S., & Farley, M. (2004). Developing a strategy for leakage man-
New Zealand. agement in water distribution systems. Water Supply, 4(3), 149–168.
Lambert, A., & Thornton, J. (2012). Pressure: Bursts relationships: https://doi.org/10.2166/ws.2004.0051
Influence of pipe materials, validation of scheme results, and impli- Tsitsifli, S., & Kanakoudis, V. (2018). Disinfection impacts to drinking
cations of extended asset life. Water Loss, 2012, 2–11. water safety: A review. Proceedings, 2(11), 603. https://doi.org/10.
Liemberger, R. (2002). Do you know how misleading the use of wrong 3390/proceedings2110603
performance indicators can be? In IWA Specialised Conference Tsitsifli, S., Kanakoudis, V., Kouziakis, C., Demetriou, G., & Lappos, S.
“Leakage Management-A Practical Approach (pp. 1–17). Lemesos, (2017). Reducing non-revenue water in urban water distribution
Cyprus. networks using DSS tools. Water Utility, 2017(16), 25–37.
Liemberger, R. (2010). Recommendation for initial non revenue water UK Water Industry Research. (2011). Best practice for the derivation of
assessment. In IWA Conference “Water Loss 2010 (Vol. 1, pp. cost curves in economic level of leakage analysis. London: UK Water
1–13). Industry Research.
Liemberger, R., Brothers, K., Lambert, A., Mckenzie, R., Rizzo, A., & UK Water Industry Research. (2013). Factors affecting background leak-
Waldron, T. (2007). Water loss performance indicators. In IWA age levels. London: UK Water Industry Research.
International Specialised Conference Water Loss 2007 (pp. Van Zyl, J. E., Lambert, A. O., & Collins, R. (2017). Realistic modeling
148–160). of leakage and intrusion flows through leak openings in pipes.
Lim, E., Savic, D., & Kapelan, Z. (2015). Development of a leakage tar- Journal of Hydraulic Engineering, 143(9), 04017030. https://doi.org/
get setting approach for South Korea based on Economic Level of 10.1061/(ASCE)HY.1943-7900.0001346
Leakage. Procedia Engineering, 119(1), 120–129. https://doi.org/10. Yousefi-Khoshqalb, E. (2019). Impact of pressure management on
1016/j.proeng.2015.08.862 the economic level of leakage in water supply networks (Master’s
Loureiro, D., Amado, C., Martins, A., Vitorino, D., Mamade, A., & thesis). Department of Civil, Water, and Environmental
Coelho, S. T. (2016). Water distribution systems flow monitoring Engineering, Shahid Beheshti University, Tehran, Iran (In persian).

You might also like