Professional Documents
Culture Documents
* Water Loss Research & Analysis Ltd, Llanrhos, Llandudno, LL30 1SL, UK, WWW.leakssuite.com
** Hydrocontrol Ltd, POBox 71044, CY3840 Lemesos, Cyprus, bcharalambous@cytanet.com.cy
*** Studio Marco Fantozzi, Via Forcella, 29 - 25064 Gussago (Brescia) Italy. marco.fantozzi@email.it
**** Aqua Libera Ltd, Ribnica 146c, 10415 Novo Cice, Croatia, jurica.kovac@mail.com
***** Institute of Applied Science, MCAST, Kordin Road; Paola, PLA9032, alex.rizzo@mcast.edu.mt
Abstract
The IWA Best Practice Water Balance and Performance Indicators were published in
1999 by the 1st IWA Water Loss Task Force, and in 2000 by the Performance Indicators
Group. The objective - to provide a long overdue standardised international approach to
the calculation of Non-Revenue Water, its components, and performance – has been
substantially achieved in an ever-increasing number of countries internationally. However,
adoption of the approach in European countries has been rather mixed; and the 2013
European Commission’s Water Blueprint’s proposed tool box approach allows each
country to choose its own performance measures for improved water loss management.
The first section of the paper highlights some common practices which are producing
misleading conclusions for water loss management performance assessment and
improvement – in particular, serious inherent problems with using %s of System Input
Volume as a performance indicator. Appendix A suggests a guideline approach to defaults
for assessed components of NRW, which could be considered as applicable to Europe.
The second section discusses Technical Minimum Real Losses - ‘How Low Could You
Go’ – assessed using IWA Water Loss Task Force equations for UARL (Unavoidable
Annual Real Losses). UARL is used in calculating the Infrastructure Leakage Index (ILI),
specifically designed to compare technical real losses management between Utilities. Key
aspects of the fundamental role of pressure management in Real Losses management
and Sustainable Economic Real Losses calculations will be in a separate future paper.
The third section summarises new developments and practical experiences since 1999
which the authors consider have clearly identified the most appropriate choices of water
loss Performance Indicators for two different purposes:
tracking progress and setting targets within a single Utility or system
comparing technical performance between Utilities with different key characteristics.
The fourth section contains updated European ILI data sets from 71 Utility systems in 12
High Income countries, and 9 Utility systems from 3 Low/Middle Income countries.
Appendix B contains a brief initial draft review of the extent to which these countries have
adopted the IWA Water Loss Specialist Group best practices relating to Water Balance
and Performance Indicators. Contributions to extend these ILI data sets to more
Countries, and to correct, update and improve Appendix B, are welcomed. Links are given
-1-
to a free access website which will hold and update international ILI data sets for Europe,
and other international regions and countries.
.
The IWA Best Practice Water Balance developed by the 1st Water Loss Task Force is
shown in Figure 1, as originally published (Hirner & Lambert, 2000; Alegre et al, 2000),
‘System Input Volume’ was defined as ‘the annual volume input to that part of the
water supply system to which the water balance calculation relates.’ This broad
definition, adopted for reasons of simplicity and flexibility, has caused some problems.
Also, Water Exported was specifically shown as a component of Billed Metered
Consumption (and also of Authorised Consumption, and System Input Volume).
Billed Billed Metered Consumption
Authorised (including water exported) Revenue
Consumption Billed Unmetered Consumption Water
Authorised
Consumption Unbilled Unbilled Metered Consumption
System Authorised
Input Consumption Unbilled Unmetered Consumption
Volume Apparent Unauthorised Consumption Non-
Losses Customer Metering Inaccuracies Revenue
Water Leakage on Transmission and/or Water
Losses Real Distribution Mains
Losses Leakage and Overflows at Storage Tanks
Leakage on Service Connections up to
point of Customer metering
Figure 1 IWA Water Balance as originally published (Hirner & Lambert 2000; Alegre et al 2000)
The first Performance Indicators Report (Alegre et al, 2000) contained an additional
diagram (Figure 2) which clearly identified ‘Water Supplied’, and numerous options for
‘System Input Volume’. Unfortunately this figure seems to be rarely referred to or used.
Exported raw water (*)
Imported raw water (*)
Imported treated
Exported treated
water (**)
water (**)
M M M
M M M M
mains
watermains
Transmission
M
intake
Distribution
Transmission
Waterintake
Treatment
Distribution
Treatment
Storage
Storage
M M
Rawwater
M M M M M M M
Water
Raw
M
Storage tanks operational
consumption and losses
Raw water consumption
M M M
Transmission
Distribution
and losses
District
Water abstracted
Water produced
Water supplied
Transmission input
Distribution input
Treatment input
metering
(*) - can be located anywhere between the water intake and the treatment
(**) - can be located anywhere downstream treatment
Figure 2: Definition of water supply system inputs and outputs (Fig. 7 of Alegre at al, 2000)
-2-
Unintended consequences and problems arising can be summarised as follows:
5 (or more) different values of ‘System Input Volume’ exist for each Utility (Fig. 3).
copies of the IWA Water Balance often omit the words ‘Water Exported’
the Standard Water Balance should have included a separate column for
‘Water Supplied’ = ‘Distribution Input’ minus ‘Water Exported’
SIV
'B'
SIV
'D'
Exported
Raw Water SIV Potable
'C' Water SIV
Imported
Imported
Potable Water Exported (metered + unmetered) 'E'
Raw Water
Billed
Authorised
SIV Consumption Billed Authorised Consumption
'A' Potable
(metered + unmetered)
Potable Water Potable
from Transmission and Distribution system
Water Produced Water
Potable Potable
Raw Water Produced and Supplied
Water Water Unbilled Authorised Consumption:
Abstracted from Imported (excluding
Produced Produced Transmission and Distribution
and Treatment Potable
Works Apparent Unauthorised consumption
Imported Non-Revenue Water
Losses Customer metering errors*
Water Water Exported)
From service reservoirs, mains
Losses Real
and service connections up to
Losses
Raw Water Raw Water point of customer metering
Abstracted Abstracted Raw Water
Consumption * Customer meter under-registration is significantly higher for
Unbilled properties supplied through customer storage tanks.
Authorised
Consumption:
Treatment
Raw Water
Losses
Figure 3: Water balance of water supply system inputs and outputs (after Fig. 7, Alegre at al, 2000)
The most meaningful results (and best practice) for quantifying and interpreting NRW
and its components will be obtained by calculating separate water balances for:
raw water systems (up to treatment works outputs)
treated water transmission mains
individual distribution systems
1.2 Best practice for estimating annual volumes of UAC and Apparent Losses
-3-
Unbilled Authorised Consumption: Water Treatment:
o % of metered treatment works potable water output
Unbilled Authorised Consumption: Distribution):
o % of Billed Authorised Metered Consumption (excluding Water Exported)
Apparent Losses:
o % of Billed Authorised Metered Consumption (excluding Water Exported)
o higher default values apply for properties having roof tanks supplied by gravity
In Appendix A, the Authors propose some maximum default values for Europe Water
Balance calculations; comments will be invited during the presentation of this paper at the
Water Loss 2014 in Vienna. These default values recognise that in some Southern
European countries around the Mediterranean (e.g. Cyprus, Malta, Southern Italy, parts of
Portugal and Spain), some or most properties have roof tanks supplied by gravity. Low
inflow rates to these tanks result in significantly higher customer meter under-registration
than in North and Central European properties with direct mains pressure and higher
inflow rates.
1.3 Assessing reliability of NRW and Real Losses calculated by Water Balance
It is an inconvenient truth, rarely acknowledged, that all calculations of NRW and Real
Losses from Water Balances are ‘best estimates’; even in fully metered systems. Any
errors in components of the calculation accumulate in the calculated Real Losses volume.
User-friendly options for calculating confidence limits of NRW and Real Losses volume
(and their Performance Indicators) have been available in some water balance software
for more than 10 years. This approach also usefully identifies which components of a
particular water balance have the greatest influence on the confidence limits of calculated
NRW and Real Losses, to prioritise further actions to reduce their uncertainty. An example
calculation of confidence limits for a simplified water balance is shown in Table 1.
Table 1: Simplified calculation example of Water Balance Confidence Limits and Priority calculation.
This type of calculation can clearly show that water balances for any European
systems with roof tanks supplied by gravity, or with large numbers of unmetered
properties (UK), are likely to have wider confidence limits for calculated real losses than
direct pressure systems. If Table 1 was a system with roof storage tanks supplied by
gravity, and CMI were 10% (not 2%), the Real Losses would be 7.4 Mm3 +/- 22%, much
reduced from 11.0 Mm3 +/- 12%, and reduction of CMI would be the second priority.
Accordingly, UK Utilities and Malta WSC use ‘bottom-up’ continuous night flow
measurements in district metered areas (adjusted with Night-Day Factors) as an
additional method of calculating Real Losses. These can then be compared with results
from ‘top-down’ water balance calculations. Some Italian Utilities also use both methods,
with confidence limits, for District Metered Areas.
-4-
Some readers may consider that the next few sub-Sections of this paper – relating to
water loss expressed as a % of System Input Volume – should be in the ‘Performance
Indicators’ Section of the paper. But it is the unanimous view of the authors that the
traditional practice of expressing NRW and its components as %s of System Input Volume
continues to perpetuate substantial yet avoidable confusion in water loss management,
and compromises the efforts of water professionals who prefer to save water rather than
to perpetuate misconceptions. The following examples are provided to justify this view.
1.4 Water Loss Volumes as % of System Input Volume – which % did you mean?
When volumes of Non-Revenue Water, Water Losses, Apparent Losses and Real
Losses are expressed as %s of System Input Volume (SIV), a single system can produce
a wide range of %s, depending on the choice of SIV used as the denominator - especially
when there are substantial exports of water - and the NRW or NRW components being
used as the numerator. The simple relationship (Aqel, 2013) is:
NRW% based on SIV = NRW% based on Water Supplied x (WS volume/SIV volume)
So if half the volume of potable water entering a system is exported from that system,
NRW or Real Losses %s based on SIV will be half of NRW% or Real Losses based on
Water Supplied. This makes %s based on SIV useless for performance comparisons
between different systems, and for comparisons between sub-systems within a Utility
where some sub-systems have exports but other sub-systems do not.
Table 2 shows an example of how calculated % NRW and its components can vary
depending upon which System Input Volume and which NRW component(s) ares selected
for the calculation. This example is for a supply and distribution system in the Southern
part of the EC, which treats a raw water of low quality, and receives imports of treated
water from an adjacent system. When a media person or politician asks a general
question such as ‘what are your % losses’, 16% 20%, 21%, 26%, 27%, 33%, 35%, 36% or
39% would all be equally valid and truthful responses. A spin doctor’s dream!
Table 2: %s by volume vary with choice of numerator (NRW component) and denominator (SIV).
NRW UAC Water Loss App. Loss Real Loss
A SOUTHERN EUROPEAN COUNTRY
% % % % %
SIV A Raw Water Abstracted 39% 7% 33% 13% 20%
SIV B Raw Water Abstracted and Imported 39% 7% 33% 13% 20%
SIV C Potable Water Produced 36% 1% 35% 14% 21%
SIV D Potable Water Produced and Imported 27% 1% 26% 10% 16%
SIV E Potable Water Supplied (exc. Water Exported) 27% 1% 26% 10% 16%
The apparent losses figure of 10% of Water Supplied (excluding exports) in Table 2
represents 14% of billed metered consumption (12.2% of true consumption), and is
actually a reasonable performance for a system with low consumption and customer roof
tanks supplied by gravity. However, in direct pressure systems in Northern Europe, some
countries with rigorous metering standards claim less than 1% customer meter under-
registration, and customer meter under-registration of more than 3.0% would be
considered high.
Irrespective of which System Input Volume is used as the basis for expressing NRW
as a %, the Apparent Losses may be expected (see Appendix A) to be significantly higher
for indirect pressure systems with roof tanks around the Mediterranean, than for direct
pressure systems in Central and Northern Europe. So all Utility Water Balances should
clearly identify if customers’ properties have roof storage tanks supplied by gravity.
-5-
Failure to identify this fact when making comparisons of NRW or Apparent Losses
within EC countries is systematically and significantly biased against systems with roof
tanks - almost all of Malta and Cyprus, and parts of Southern Italy, Portugal, Spain..
500 lit/conn/d
50%
700 lit/conn/d
40%
1000 lit/conn/d
30%
1500 lit/conn/d
20% 2000 lit/conn/d
0%
0 50 100 150 200 250 300 350 400 450 500
Real losses litres/connection/day
Figure 4: How differences in consumption distort perception of Real Losses when %s are used
Figure 4 shows that arbitrary targets (e.g. 10% Real Losses for all systems) are almost
meaningless in terms of technical performance comparisons for management of Real
Losses (and NRW). The green star is a rural Utility, and the red star a major city, both in
the same European country, each with Real Losses of 8.8% of System Input Volume:
The ‘green star’ rural Utility, with 470 litres/service/connection consumption, has had
to achieve Real Losses of 49 litres/service connection/day (ILI = 2) to achieve 8.8%
But because the ‘red star’ city has 4200 litres/conn/day consumption, it can have real
losses of 426 litres/service connection/day (ILI = 18), or almost 9 times as much
leakage as the ‘green star’ city, and still achieve 8.8%!.
-6-
From year to year, leakage per connection may rise, stay constant or fall; and total
consumption per connection may rise, stay constant or fall. So changes in %s of Real
Losses (and % NRW) expressed as a % of SIV are influenced by changes in both
leakage and consumption over the same period. The red dashed vertical line in Figure 4
shows that for any particular volume of Real Losses in litres/service connection:
if consumption reduces, Real Losses as a % of Water Supplied increases, giving the
impression that Real Losses volume has increased, when in reality it has not changed.
if consumption increases, Real Losses as a % of Water Supplied reduces, giving the
the impression that Real Losses has reduced, when in reality it has not changed.
The problem is illustrated in Table 3. Suppose that in a base year, Real Losses are 40
Ml/d (20% of Water Supplied) and the target for Real Losses is calculated as an Economic
Leakage Level (ELL) or Sustainable Economic Leakage Level (SELL) as being 32 Ml/d. If
this target is expressed as a % (16%) of Water Supplied, and it takes several years to
achieve Real Losses of 32 Ml/d, then changes in consumption during these years can
significantly influence the perception as to whether the target has been attained, or not.
Table 3: Real Losses targets expressed as %s of a measure of SIV corrupts performance tracking
A Consumption reduces by greater % than real losses Real Losses appear to increase
B Consumption reduces by same % as real losses Real Losses seem to be unchanged
C Consumption reduces by smaller % than real losses Real Losses reduction under-estimated
D Consumption remains the same as base year Real Losses slightly under-estimated
E Consumption increases compared to base year Real Losses reduction over-estimated
So continuing widespread use of %s in Europe for tracking changes in NRW and Real
Losses is not just a theoretical problem, as the results cannot avoid being compromised
by changes in consumption. In recent years, many Utilities throughout Europe have
promoted demand management (low flush toilets, taps and shower heads etc) and
increased prices of water, and have achieved the desired outcome of reducing
consumption substantially. In German utilities with distribution systems already operating
at low leakage levels close to the technically achievable minimum, Real Losses expressed
as a % have therefore inevitably been increasing as consumption reduces (Tennhardt,
2012; Water Loss Detectives, 2012).
http://www.leakssuite.com/leakage-tracking-percentages/ discusses the two above
examples from Germany, and also shows an Australian example of Outcome B in Table 4,
where during a 7 year drought a large city reduced both leakage and consumption
volumes by around 50%, without any significant change showing in Real Losses as % of
System Input Volume. Fortunately the Australian WSAA had accepted IWA
recommendations of 1999/2000 and ceased to use or publish %s in National Statistics in
2003. Sadly, the Regulator for that particular Australian State has still not been recognised
the fallibility of using %s!
-7-
In practice, it is preferable that, once Real Losses targets have been calculated in
volume per year, they are converted to a ‘per service connection’ or ‘per km of mains’
basis (whichever is most appropriate to a particular country), as that allows for growth in
the infrastructure over time, and permits simple tracking of year on year changes in
multiple sub-systems, using graphs such as Figure 6 or Figure 7 (in Section 2 of this
paper).
There is only perhaps only one situation where Real Losses (or NRW) volume as a %
of System Input Volume is a perfect performance indicator. That is for proving that H.L.
Mencken’s famous quote, below, is 100% correct
‘For every complex problem, there is a solution that is simple, neat and wrong’ .
In addition to developing a Best Practice Water Balance, the 1st Water Loss Task Force
(Lambert et al 1999) was required ‘to review Performance Indicators (PIs) for international
comparisons of losses in water supply systems’.
After identifying problems with %s, described in Sections 1.4 to 1.6 of this paper, the Task
Force concluded that ‘per service connection’ and ‘per km of mains’ were the best of the
traditional performance indicators, as they represented the two parts of the distribution
infrastructure where almost all leakage occurred. But neither of these two PIs took
account of average pressure, density of connections per km of mains, and average length
of service connection between main and customer meter, so they were not suitable for
comparisons between systems.
The Task Force therefore developed a performance indicator (Infrastructure Leakage
Index ILI) specifically designed for technical comparisons of Real Losses from systems
with different key infrastructure and pressure characteristics. The starting point is a
formula for technical minimum or ‘Unavoidable Annual Real Losses’ (UARL) based on an
auditable component analysis of real losses for well maintained infrastructure in good
condition. The Infrastructure Leakage Index (ILI) could then be calculated as the non-
dimensional ratio of Current Annual Real Losses/UARL. The ILI could then be used for
‘level playing field’ technical comparisons of real losses management between different
systems with different key characteristics.
-8-
where Lp is the average length of the whole of the service connection (main to customer
meter) in metres. The coefficient of 0.8 for Nc in Equation 1 has been reduced to 0.7 in
equation (2), assuming approximately 4 metres per service connection, from main to
property line. Equations (1) and (2) produce almost identical results for UARL.
Understandably, the existence of two different equations for UARL caused users some
confusion, and in the 2nd Edition of the Performance Indicators Report (2007), the UARL
formula was changed back to Equation (3), similar to Equation (1) but in which Lp is the
average length of the service connection from the property line to the customer meter) in
metres (see also Figure 5):
UARL equations (1), (2) and (3) can easily be converted to a wide variety of units
(annual, daily, hourly; in m3 or litres; per km mains or per service connection). If UARL in
m3/year is required for direct comparisons with CARL in m3/year from a European Water
Balance, then equations (1) and (2) can be converted to:
UARL (m3/year) = (6.57 x Lm (km) + 0.292 x Nc + 9.13 x Lp (km)) x P ……..(1a)
UARL (m3/year) = (6.57 x Lm (km) + 0.256 x Nc + 9.13 x Lt (km)) x P ……..(2a)
where Lp in equation 1a is total length (km) of service connections, property line to meter,
and Lt in equation 2a is total length (km) of service connections, main to meter
The UARL formula was originally developed for comparing performance of whole Utilities.
In 1999, international testing on 27 systems in 18 countries identified only one system (in
Netherlands) with an ILI less than 1.0 (0.7), with an average pressure of 35 metres, where
almost all leaks appeared to surface, and there was no need for active leakage control.
Since 1999, the UARL formula has proved to be robust in predicting ‘how low could you
go’ with Real Losses in best-performing Utilities. Internationally, comparatively few Utilities
have been able to achieve Real Losses equal to their predicted UARL (ILI close to 1.0)
except in special circumstances (e.g. Netherlands ILI 0.7) which fall outside the original
assumptions and the current limits of application of the UARL formula, see Section 2.4.
-9-
2.3 UARL in Litres/connection/day and m3/km mains/day
For systems where customer meters are located at the property line (Lp=0 metres per
service connection). , the influence of density of connections and average pressure is
shown in Fig. 6 (UARL in litres/conn/day) and Fig. 7 (UARL in m3/km mains/day). Similar
graphs can be created for any other value of Lp.
Figures 6 and 7 clearly show that it is not realistic to expect all Utilities to achieve
the same technical real losses management performance on a ‘per service connection’ or
‘per km of mains’ basis. Each system has its own specific UARL which, for different
combinations of density of connections and pressure, can vary between 92 and 23
litres/connection/day (Figure 6) or between 0.5 and 5.5 m3/km/day (Figure 7). So these
performance indicators are not recommended for making technical performance
comparisons between systems.
- 10 -
Figure 7: UARL in m3/km mains/day, for customer meters at property line
The original IWA Task Force recommendation to use ‘per connection’ if ≥ 20 conns/km,
and ‘per km’ if < 20 conns/km, arose because more than half of the UARL occurs on
connections when connection density exceeds 20 per km (as seen in Fig 7), and this is a
useful Guideline to follow. But as 20 to 40 conns per km is a tricky region, either option is
reasonable, so preference should be given to whichever of the two PIs has been used
traditionally in any particular country.
However, %s based on any definition of System Input Volume (including Treated Water
Supplied) should definitely not be used for tracking progress in NRW or real loss
management within individual systems, for reasons explained in Section 1.6, Figure 4 and
Tables 3 and 4.
UARL and ILI were originally developed for application to whole Utilities with mixed
pipe materials, using the Background and Bursts Estimates concept for component
analysis of real losses, and the FAVAD concept for pressure:leak flow rate relationships.
Components for background leakage, reported and unreported leaks and bursts are
based on clear and auditable assumptions for frequencies, flow rates and durations of
leaks, at 50 metres pressure. These are shown in Table 1 of Lambert (2009).
Lower limits for density of connections, average pressure and system size have
always been recommended for application of the UARL formula. Table 5 (from Lambert,
2009) shows changes in these limits between 1999 and 2009 which reflect the increasing
international use of ILI and Snapshot ILI (derived from night flows) for small systems and
DMAs. The graph referred to in Table 5 is Figure 2 in Lambert (2009), which can be
obtained through the weblink
http://173.254.28.127/~leakssui/wp-content/uploads/2012/11/2009_LambertWaterlossCapetown-2009J.pdf
- 11 -
Table 5: Changes in limits of application of UARL formula, 1999 to 2009
Parameter Limits Lambert et Lambert & Liemberger Lambert
al, AQUA McKenzie & McKenzie
1999 2001 2005 2009
Density of Minimum 20 20 Removed No lower limit
Connections/km Maximum 100 Removed No upper limit
Average Minimum 20 25 25 See Graph
Pressures (m) Maximum See Graph
System Size Minimum Not stated Nc > 5000 Nc > 3000 Nc + 20 x Lm > 3000
Source: Lambert (2009)
Table 6, from the same paper shows the summarised components of the UARL
formula at 50 metres pressure, together with the FAVAD N1 exponents which should be
used to predict how leak flow rates of UARL components vary with average pressure.
Since 1999, UARL and ILI have been calculated for thousands of systems
internationally. At first, most practitioners considered that UARL and an ILI of 1 were
simply not achievable, but at least 50 of the best-performing medium and large Utilities
internationally are now achieving ILIs close to 1.
Most Utilities in Europe with more than several thousand connections, and average
pressures of 35 metres or more, would find it extremely challenging to try to achieve and
maintain an ILI close to 1.0. Practical guidance developed in Australia for WSAA (2009) is
to check any calculated ILI less than 0.7, and it is remarkable how often this simple quality
control quickly identifies one or more of the following data problems:
Systematic and random errors in the Water Balance affect assessed volumes of NRW
and Real Losses (see Table 1), and all related performance indicators (not only ILI):
o Bulk metering errors (under-registration of true volume)
o Meter Lag errors (bulk and customer meter readings not synchronised)
o Billing systems designed for revenue collection, not technical data retrieval
o Over-estimates of unmetered consumption
o At low leakage, confidence limits for Real Losses are usually at least +/- 10%
o Confidence limits widen if there are significant volumes of water exported, or
high estimates of apparent losses (especially with customer storage tanks)
- 12 -
Errors associated with other parameters used to calculate UARL
o Number of service connections over-estimated by using number of properties
(higher) as a substitute figure
o Average pressure not systematically calculated, and over-estimated
o assumptions used in UARL are not all valid for system under consideration
In systems with high burst frequencies, the standard UARL formula can be used for
smaller systems than those in Table 3, as ILIs will be greater than 1.0. However, when the
Austrian OVGW adopted the ILI as the decisive PI for Real Losses (OVGW 63, 2009),
significant numbers of very small Utilities (most with fewer than 3000 service connections
and relatively new infrastructure) began to report ILIs less than 0.7.
A research study (Lambert and Koelbl, 2014) is currently investigating how component
analysis, FAVAD and the latest prediction methods for pressure:bursts relationships might
be used to predict UARL for small systems (equivalent to District Metered Areas) with
good infrastructure. This may result in a further update of Table 4, or possibly a practical
prediction method in which corrections are applied to the standard UARL formula
depending upon combinations of circumstances in the small systems (notably average
pressure, pipe materials and number of service connections).
The 1st Edition of ‘Performance Indicators for Water Supply Systems’ (Alegre et al,
2000), proposed 133 Performance Indicators covering a wide range of Water Utility
functions. Table 7 shows the small subset of NRW and Water Loss Performance
Indicators, developed by the 1st IWA Water Loss Task Force (Lambert et al, 1999), which
were included in the 1st Edition of the PIs Report.
Table 7: Recommended performance Indicators, for different purposes (IWA, 2000)
Note: Each indicator was also assigned a Level of Importance (L1 to L3), and a Code (e.g. Op23), but
these are not shown in Table 7, as Levels of Importance were removed and some of the Code numbers
were changed in the 2nd Edition (Alegre et al, 2007).
- 13 -
3.2 Performance Indicators Report, 2nd Edition (2006)
Table 8 summarises the NRW and Water Loss Performance Indicators recommended in
the 2nd Edition of the PIs Manual, after testing by the IWA Performance Indicators Task
Force with 70 volunteer Undertakings in Europe, Asia-Pacific, Africa and South America.
The IWA Water Loss Task Force, which had been assisting several other countries
(Australian WSAA, New Zealand NZWWA, North America AWWA) and numerous Utilities
to implement the 1st Edition IWA Water Balance and Performance indicators, had no
significant input to the 2nd Edition. Consequently some differences of emphasis began to
arise as to what were considered to be ‘best practice’ IWA Water Loss Performance
Indicators (Liemberger et al, 2007). In North America, Fanner et al (2007) identified that
some Water Loss PIs were more suitable for target setting and monitoring of progress
within individual Utilities, whilst others were more suitable for comparing technical water
loss management performance between Utilities with different key characteristics.
nd
Table 8: Recommended performance Indicators, 2 Edition (2006), IWA Performance Indicators Group
- 14 -
3.3 IWA Performance and Benchmarking Assessment Specialist Group Initiatives
In 2008 a new IWA benchmarking initiative (Cabrera, 2008) split benchmarking into two:
Metric benchmarking: compares PIs of utilities with different characteristics
Process benchmarking: identify and adapt best practices to improve performance
In 2010 the IWA Benchmarking and Performance Assessment Group (Cabrera et al,
2010) implemented an evolution which replaced ‘Metric’ and ‘Process’ Benchmarking with
Performance Assessment and Performance Improvement, see Figure 8.
METRIC PROCESS
Source, Cabrera et al 2010, with overlay linking to former ‘Metric’ and ‘Process’
Figure 8: The new IWA Benchmarking Framework :
These developments greatly simplified the selection of individual NRW and Water Loss
Performance Indicators which are most appropriate for either:
Target setting and monitoring progress in individual systems and sub-systems,
or
Technical performance comparisons between systems with/without customer
storage tanks, and having different connection densities, meter locations and
operating pressures
3.4 Target Setting and Monitoring within Individual Systems and Sub-systems
Table 9 shows Performance Indicators (and variables and context information) which are
currently considered by the authors of this paper as being the most appropriate for
practical target setting and monitoring of progress in the management of NRW
components for individual systems and sub-systems within an individual Utility.
Within each individual system, connection density and customer meter location on the
service connection are likely to be reasonably consistent over time. Speed and quality of
- 15 -
repairs, active leakage control and pressure management are the main short to medium
term options used to manage real losses.
Table 9: The Authors’ recommended PIs, variables and context data for target setting and annual
progress monitoring in management of NRW components of individual distribution systems
Recommended Water Loss PIs for Target Setting and Monitoring Progress in an Individual Distribution System
Component of Water Balance, and Parameter used as PI
Prime
additional Variables and Context data for
data Connections/km mains Comments
short to medium term real loss
units
management Low Medium/high
Potable Water Supplied PWS m3/year m3/km/yr m3/conn/yr PWS excludes all Water Exported
3 3 3 Countries with strong traditional
Billed Authorised Metered BMC m /year m /km/yr m /conn/yr
Consumption BAC practice of using per km mains, per
Unmetered BUC m3/year m3/km/yr m3/conn/yr service connection, or per property,
Non – Revenue Water NRW 3
m /year 3
m /km/yr m3/conn/yr or per day (not per year) should
continue to use that measure, for
Unbilled Authorised Consumption UAC m3/year m3/km/yr m3/conn/yr
familiarity and continuity. Or decide
Apparent Losses AL m3/year m3/km/yr m3/conn/yr based on average connection
3 3
Real Losses RL m /year m /km/yr m3/conn/yr density for the Utility's systems.
3 3
Total Consumption TC = PWS -RL m /year m /km/yr m3/conn/yr Use of %s is NOT recommended
3
UAC + Apparent Losses AL m /year % of BMC % of BMC These measures help to set targets
Economic Annual Real Losses EARL m3/year m3/km/yr m3/conn/yr and interpret performance in the
Unavoidable Annual Real Losses UARL m3/year m3/km/yr m3/conn/yr management of NRW components
Infrastructure Leakage Index = RL/UARL Annual ILI and/or Snapshot ILI Use Snapshot ILI for night flows
Location + Repair Mains leaks Days or Hours Average time from awareness of
time Service leaks Days individual leaks to repair/shutoff
Leaks on Mains No/year Number per 100 km/yr Reported, and unreported leaks
found by ALC. Exclude small leaks
Leaks on Service Connections No/year Number per 1000 conns/yr on valves, meters and stop taps.
Active Leakage Control Interventions % of total mains length in year % subject to ALC interventions
3 3
Rate of Rise of Unreported Leakage m /year m per year in 1 year For ALC Economic Intervention
Active Pressure Management % of total mains length % subject to pressure management
Average pressure metres When system pressurised
% of time system is pressurised % of year Only if intermittent supply
- 16 -
Frequency of leaks on mains and services: always show these separately (as well as
combined, per km, if that is traditional practice) to highlight which part of infrastructure
has problems. Separate mains and services figures are needed for predictions of burst
reduction from pressure management
Active Leakage Control: ALC installations are only effective if interventions are
frequent enough; rate of rise of unreported leakage allows economic frequency of
intervention to be quickly assessed, depending on marginal cost of water and
intervention cost
Pressure management:: reduction of excess pressures and control over pressure is
now recognised as the fundamental foundation of effective real loss management,
influencing leak flow rates, leak frequency, rate of rise, asset life etc.
Table 9 shows Performance Indicators (and variables and context information) which are
currently considered by the authors of this paper as being most appropriate for technical
performance comparisons of the management of NRW components, between different
systems with different key characteristics (comparing apples with pears).
The European Commission’s Water Blueprint (2013), does not advocate a ‘one-size-
fits-all’ straitjacket but proposes a tool box that member states can use to improve water
management. This leaves each country able to choose its own performance measures for
water loss management. Provided that potentially misleading %s of System Input Volume
are not used, this approach has the benefit of continuity and familiarity for using m3/km
mains per year, litres/connection/day or litres/property/day to set targets and track
performance of individual systems in different countries. However, none of these
measures are suitable for rational comparisons of technical performance between Utilities.
Superficial comparisons of NRW and Real Losses using %s of System Input Volume
systematically favour large Utilities with high consumption (see Fig 4) and direct pressure,
and numerous different % figures can be quoted depending upon which definition of
System Input Volume is used (see Figure 3 and Table 2). Also, changes in NRW and Real
Losses expressed as %s of SIV only represent true % changes if consumption is
constant. However, with increasing unit charges for water and active measures to reduce
consumption, European Utilities with low Real Losses have inevitably experienced year-
on-year increases in % NRW and Real Losses, as true % reductions in total consumption
have been reducing faster that true % reductions in NRW and Real Losses (see Tables 2
and 3).
- 17 -
o express UAC as % of Billed Metered Authorised Consumption (excluding
Water Exported)
o This will normally be less than 2% for direct pressure systems with reliable
metering, but more than 5% for systems with roof tanks supplied by gravity
(see Appendix B)
Real Losses represent treated water that has been lost from distribution systems by
leakage and overflows; the part of this in excess of the Technical Minimum Real
Losses (UARL) represents a potentially recoverable water resource
o calculate Unavoidable Annual Real Losses UARL at current average pressure
o express Current Annual Real Losses (CARL) as the ratio CARL/UARL = ILI
o ILI is likely to can range from close to 1.0 for some systems in High Income
Countries to as much as 30 for some systems in low income countries
o Use World Bank Institute Banding System to broadly classify ILI performance,
and identify appropriate priority actions for individual systems
Use Context Information in Table 9 and Water Balance, together with items below,
for further interpretation of relative performance and management opportunities
o Number of billed properties (residential and non-residential) and populaltion
o Infrastructure information:(mains length, number of service connections,
average length of service connection, main to meter.
o Average pressure (metres)
Of the 27 system that were originally used to test the UARL equation and the ILI
concept (Lambert et al 1999), 20 were from European countries (Denmark, England,
Finland, France, Germany, Gibraltar, Greece, Iceland, Malta, Netherlands, Northern
Ireland, Switzerland, Spain, Sweden and Wales). Many of these were from large cities.
Some ILIs for Utilities in Cyprus and Italy were then added to most of the 1999 ILIs to
produce a 2005 European set of 22 ILIs from 12 European Countries (Figure 9).
The Banding system shown at the right hand side of Figure 9 was introduced by the
World Bank Institute into its NRW Training Modules in 2005. It uses a matrix for assessing
Real Losses management performance, based on real losses in volume/service
connection/day for a range of average operating pressures, and classified into Bands A to
D. The targets assume that customer meters are located at the property line boundary,
with an average connection density of around 40 per km mains.
Bands A to D in the WBI target matrix were in practice calculated from an equivalent
range of ILIs, which can be applied to a wider range of connection densities and customer
meter locations. Band limits in terms of ILIs, general descriptions of each Band, and
appropriate recommended actions are shown in Tables 10 and 11.
Low and Middle Income (LAMIC) countries receive double the ILI range limits
compared to Higher Income Countries (HIC), based on whether average per capita
income is above or below $12k per person per year - see
http://www.iwahq.org/contentsuite/upload/iwa/all/2012%20HIC-LIC%20Country%20Listing%281%29.pdf
- 18 -
Figure 9: European data set of 22 ILIs from 12 Countries, 2005
Table 10: WBI General Descriptions of Real Loss Performance Categories (2005)
Table 11: Recommendations for Appropriate Actions for WBI Bands (2005)
WBI Recommendations for BANDS A B C D
Investigate pressure management options Yes Yes Yes
Investigate speed and quality of repairs Yes Yes Yes
Check economic intervention frequency Yes Yes
Introduce/improve active leakage control Yes Yes Yes
Identify options for improved maintenance Yes Yes
- 19 -
By 2010, many Utilities internationally had achieved ILIs close to 1.0. In Australia
during the multi-year millennium drought so many Utilities achieved ILIs in Band A that it
was considered appropriate to split into Bands A1 and A2. The splitting of WBI higher
Bands into B1, B2, C1, C2, D1, D2 has also been found to be helpful in a LAMIC country
(Malaysia) outside Europe, as it gives incentive for Utilities to celebrate achievements in
moving from, say C2 to C1 (which otherwise represents a very long journey, reducing
from the top of C (ILI = 8) to the bottom of C (ILI < 4).
8 to < 12 4 to < 6 C1 Poor leakage record; tolerable only if water is plentiful and cheap;
even then, analyze level and nature of leakage and intensify
12 to < 16 6 to < 8 C2 leakage reduction efforts
ILI data from several countries, both European and International, is now published, but
is not yet currently available in one easily accessible location, free of charge to anyone
interested. Accordingly, one of the authors has now created an updated set of 71 ILIs from
12 High Income European Countries (Figure 10) and 12 Utilities in 3 Low/Middle Income
European Countries (Figure 11), using both attributed and anonymous Utility data. The ILI
data for High Income countries has been quality controlled, to a greater or lesser extent,
by National organisations or experienced IWA Water Loss Specialist Group experts.
From early April 2014, this ILI data set, together with quality controlled ILI data sets
from other Countries and World regions (Australia (NWC), Balkans, Canada, North
America (AWWA) ……) will be freely available for downloading from
http://www.leakssuite.com/global-ilis/ .
A brief but only limited review of additional relevant information on Water Balances and
performance indicators currently used in the European Countries which provided ILI data
is provided in Appendix B. The authors would welcome any feedback on errors or
omissions in these reviews, as this is important context information which could also be
located, updated and made freely available to European and International leakage
- 20 -
specialists through the website. Any further contributions of good quality ILIs and
associated information from any European country will of course be welcomed.
ILIs for 71 Water Utilities in 12 European High Income Countries , circa 2012 : data set at 28 Feb 2014
20.0
19.0
18.0
17.0
16.0
15.0
WBI
14.0 Band
13.0 D
Infrastructure Leakage Index
12.0
11.0
10.0
9.0
8.0
7.0
WBI
6.0 Band
5.0 C
4.0
3.0 WBI
Band B
2.0
1.0 WBI
Band A
0.0
Germany Land…
Italy IREN A
Italy (An)
Italy IREN I
Italy IREN J
Croatia (An)
Croatia (An)
Croatia (An)
Croatia (An)
Croatia (An)
Italy IREN F
Italy IREN E
Croatia (An)
Croatia (An)
Croatia (An)
Croatia (An)
Croatia (An)
Croatia (An)
Croatia (An)
Italy IREN L
Croatia (An)
Croatia (An)
Croatia (An)
Croatia (An)
Croatia (An)
Croatia (An)
Croatia (An)
Croatia (An)
Croatia (An)
Croatia (An)
Italy IREN B
Italy IREN C
Italy IREN K
Mal ta (WSC)
Italy IREN D
Italy IREN G
Italy IREN H
Italy IREN N
Belgium DeW g_B
Italy IREN M
France (An)
France (An)
France (An)
France (An)
Eng/W ales (An)
Eng/W ales (An)
Gozo (W SC)
Spain (An)
Cyprus (An)
Austria (An)
Belgium DeW g A
Austria (An)
Austria (An)
Austria (An)
Cyprus (An)
Austria (An)
Austria (An)
Austria (An)
Portugal (An)
Germany Land 2
Germany Land 3
Germany Land 4
Belgium DeW g C
Figure 10: ILIs for 71 Water Utilities in 12 European High Income Countries
Austria, Belgium, Croatia, Cyprus, England/Wales, France, Germany, Italy, Malta, Portugal, Spain, Switzerland
Figure 11 shows ILI data for 12 Utilities in 3 European ‘LAMIC’ Countries; the WBI
Band limits for ‘LAMIC’ countries are twice the Band Limits for Higher Income Countries.
ILIs for 12 Water Utilities in 3 European Low/Middle Income Countries, circa 2012: Data set as at 28 Feb 2014
21.0
20.0 WBI
19.0 Band
18.0
D1
16<ILI <32
17.0
16.0
15.0 WBI
14.0 Band
Infrastructure Leakage Index
13.0 C2
12.0
11.0 WBI
10.0 Band
9.0 C1
8.0
WBI
7.0
Band B2
6.0
WBI
5.0
Band B1
4.0
3.0
WBI
Band A2
2.0
1.0
WBI
Band A1
0.0
Bosnia HG Serbia (An) Bosnia HG Serbia (An) Serbia (An) Bosnia HG Serbia (An) Bulgaria (An) Serbia (An) Serbia (An) Bosnia HG Serbia (An)
(An) (An) (An)
Figure 11: ILIs for 12 Water Utilities in 3 European Low Income Countries
Bosnia, Bulgaria, Serbia
- 21 -
Some quality control has been applied to the LAMIC Water Balances and PI
calculations, but it is usually unrealistic to expect a high degree of reliability, without major
investments in bulk and customer metering and billing systems.
Acknowledgements
The authors wish to thank the many Water Loss Specialist Group members who have
contributed to the development and implementation of the IWA Best Practice Performance
Indicators; the many Utilities who have contributed ILI data to the European data set; and
all those practitioners who seek to continue to improve international understanding and
best practice for measurement and management of Non-Revenue Water and its
components.
Hirner, W and Lambert, A. (2000) Losses from Water Supply Systems: Standard Terminology and
Recommended Performance Measures. IWA Blue Pages, October 2000
Alegre H, Hirner W, Baptisata J.M., Parena R (2000). Performance Indicators for Water Supply Systems. IWA
Publishing ‘Manuals of Best Practice’ series, ISBN 1 900222 272, July 2000
Alegre H, et al. (2006). Performance Indicators for Water Supply Services, Second Edition. IWA Publishing
‘Manuals of Best Practice’ series, ISBN 1843390515, July 2000
Fanner P.V., et al (2007). Leakage Management Technologies. AWWARF Project Report 2928
Cabrera E Jr, (2008). ‘Benchmarking in the Water Industry: a mature approach. Water 21, August 2008, p. 64
Liemberger et al, (2007). Water Loss Performance Indicators. IWA International Specialised Conference
‘Water Loss 2007’, Bucharest, September 2007. Conference Proceedings (3 vols) ISBN 978-973-7681-
24-9; Session B2 of Volume 1
http://173.254.28.127/~leakssui/wp-content/uploads/2012/11/2007_LiembergerBrothers-et-al-
IWABucharest-2007T.pdf
Australian Government, National Water Commission: National Performance Report, 2010-11, Urban Water
Utilities.
Lambert, A (2009) Ten Years Experience in using the UARL formula to calculate Infrastructure Leakage
Index. IWA Specialist Conference ‘Waterloss 2009’,Cape Town, South Africa, March 2009
http://173.254.28.127/~leakssui/wp-content/uploads/2012/11/2009_LambertWaterlossCapetown-2009J.pdf
- 22 -
Austrian Association for Gas and Water (2009). Guideline W63 ‘Water losses in water supply systems’ OVGW
W63 September 2009
Cabrera et al (2010): Forget ‘metric’ and ‘process’, benchmarking is being reborn: the new IWA benchmarking
framework. IWa Water Utility 21, December 2010.,
Tennhardt L (2012). Realistic Balance of Water Losses and the Practical Use of Water Loss Indexes. Energie:
Wasser-Praxis, 10/2012, pages 34 to 41
Water Loss Detectives, 2012: ‘Operational Efficiencies from permanent leakage monitoring’. Article by F
Tantsky, Albstadtwerke. May 2012 Issue, of magazine published by AQUATIM, Romania
(alin.anchidin@gmail.com).
Aqel S (2013) NRW Comparison equations (Power Point Presentation): personal correpsondence
- 23 -
Appendix A:
If a water utility uses values greater than the above defaults, sufficient data should be
provided to satisfy an auditor as to the accuracy of the values used. As a minimum, for
under-registration of retail meters, the following must be provided:
The water utility should be consistent across reporting years in calculating its Current
Annual Real Losses and, where appropriate, have supporting documentation to verify
assumptions for the purpose of auditing.
- 24 -
Appendix B:
Table: Descriptions of Austrian OVGW categories Performance Bands A to D, from OVGW 63 (2009)
Austria has around 5500 Water Utilities, many with fewer than 3000 service
connections, which is the 2009 international lower guideline limit for ILI calculations (see
Table 5 of paper). During the OVGW trials to evaluate ILI, ILIs less than 1.0 were reported
from the smaller systems. Some of these were later identified as having systematic water
balance errors (often unmetered public fountains) but it’s clear that ILIs less than 1 are
occurring in some small Austrian systems. A current research study (Lambert and Koelbl,
2014), applying component analysis and the latest pressure:bursts prediction methods to
check if low ILIs in small systems are consistent with UARL concepts, is now in an
advanced stage. This is likely to result in a further update of Table 4.
Austrian OVGW now has a national unpublished data set of more than 100 ILIs. The
data set of 7 Austrian ILIs in Figure 10 of the paper were taken from a study by Koelbl
(2007), and appear to roughly represent the range (0.6 to 6) of ILIs in Austrian systems
with more than 3000 service connections, although it is expected that many of the smaller
Austrian systems will have ILIs close to or somewhat below 1.0.
Bosnia and Herzgovina: Water utilities in Bosnia and Herzegovina are still struggling
with numerous problems related with political and economic issues. Despite existing
problems some Utilities have shown progress, with international recognition of achieved
results by Gracanica water utility. A bright point is the activity of the Hydro-Engineering
Institute Sarajevo which regularly uses the IWA methodology and PIs in water loss control
- 25 -
projects. Jurica Kovac has collated ILI data from 4 Utilities as Bosnia’s contribution to the
2014 to the LAMIC European data set.
Belgium : most Utilities use the IWA Water Balance and technical measures of the IWA
practical approach to NRW management such as sectorisation, active leakage control and
pressure management. However, contrary to IWA best practice recommendations, %s by
volume are still predominantly used for Real Losses comparisons and target setting.
The Belgian contributions to the ILI data are from De Watergroep, a large Utility serving
numerous rural communities in Flanders (2.8 million customers, 1.2 million connections,
31000 km mains length). This is an excellent example of how use of %s by volume can
adversely distort the true perception of real losses management performance of Utilities
with low consumption. With a high marginal price of water and one of the lowest
consumption rates in Europe (300 litres/service connection/day), NRW of 20% actually
hides a very creditable average ILI of 1.55, in WBI Band A. ILIs for De Watergroep’s 4
large sub-systems are included in Figure 10.
Bulgaria adopted the IWA Water Balance in 2005, but not the IWA performance
indicators. However, some Water Companies have started to calculate ILIs using free
Water Balance and PI software translated into Bulgarian language (CheckCalcs or
EasyCalcs), and are reporting these at Bulgarian Water Association annual National
Conferences. Although Bulgarian legislation requires water balances to be carried out, at
present system input and output are mostly not being measured, but estimated.
However, one active city (Dryanovo) is reporting success in using ILI to monitor
improvements in real loss management in small Zones (70 to 200 services, 350 to 1000
service connections) with high burst frequencies. Dr Atanas Paskalev of AQUApartner
arranged for an English translation of a paper by Indzhov et al (2013), which showed
recently reported Zone ILIs are in the range 2.7 to 33, with weighted system average ILI
13.5; this is Bulgaria’s contribution to the LAMIC data set. Dryanovo is a good example to
show that the lower limits of application of ILI need not be applied to systems with poor
infrastructure, high burst frequencies, high background leakage, and relatively high ILIs.
A major Conference in 2007 introduced the IWA approach, supported by free IWA
Water Balance and PI software in Croatian language. In 2009 the Association of Water
Utilities encouraged water utilities to start using the IWA methodology, and from July 2010
Croatia introduced new legislation (planned to be implemented in 2015) which included
performance indicators based upon IWA Water Balance and Performance Indicators, and
concessions payments for the water extracted by the water utilities, with 4 price categories
calculated using a water losses coefficient based on WBI bands and the ILI. However,
- 26 -
following changes in senior personnel at the Government Agency Croation Waters, a
uniform high fixed unit charge has been implemented which should stimulate more
leakage control by all Utilities.
The Croatian government plan to unite 150 small and mid-size utilities into 21 large
utilities, starting in 2014. Another interesting development is a dedicated German/Croatian
Training and Competence Centre for water utilities established in the town of Karlovac,
where pilot courses, including Water Loss topics based on IWA best practice
methodology, have been implemented. In 2014 this Training Centre initiative has grown to
expand its influence on the whole region. Jurica Kovac has collated ILI data from 23
Utilities as Croatia’s contribution to the 2014 HIC European data set.
Cyprus almost all customers have roof storage tanks on this Mediterranean island. One
of the larger Utilities was an early and enthusiastic user of the IWA practical approach,
and achieved an ILI below 2.0 by 2007, before a 3-year drought period involving
intermittent supply caused substantial increases in bursts and permanent damage to the
distribution system in term of increased background leakage. ILIs for two of the three
larger Utilities have been contributed anonymously to the international data set.
In February 2014 the Water Development Department of the Ministry of Agriculture,
Natural Resources and Environment of the Republic of Cyprus, with the collaboration of
the Water Boards, Union of Municipalities and the University of Cyprus, published a “Code
of Best Practice for the Management and Operation of Water Distribution Networks” which
adopts the IWA WLSG strategies and methodologies.
*Charalambous B.(2011) ‘The effects of Intermittent Supply on Water Distribution
Networks’, IWA Efficient 2011 Conference, Dead Sea, Jordan, Proceedings. Free copy:
http://173.254.28.127/~leakssui/wp-content/uploads/2012/11/2011_Charalambous-2011J1.pdf
France The AGHTM proposed a standard water balance and performance indicators for
France in 1990, and the extent to which the IWA standard Water Balance (2000) has
replaced it is not clear to the authors. Use of %s in the form of Rendement (% of system
input volume delivered) and Linear Leakage Index LLI (Indice Lineare des Pertes, m3/km
mains/day) are the most frequently used PIs for water losses (apparent plus real), but with
direct pressure systems apparent losses are normally quite small. LLI values are then
usually converted into some kind of descriptive Performance Classification e.g.
‘Worrying/Mediocre.Almost Satisfactory/Satisfactory’ as in the FFNCR Classification of
2003
.
Table: Performance Classification according to FNCCR proposal (2003)
- 27 -
Class Limit Discontinuities: Examples: FNCCR, Laboratoire Gea (per km, per conn)
Source: Fantozzi, Lambert and Liemberger: Some Examples of European Water Loss Targets,
and the Law of Unintended Consequences. IWA Specialised Conference ‘Water Loss 2010’,Sao
Paolo, Brazil, June 2010
http://173.254.28.127/~leakssui/wp-content/uploads/2012/11/2010_FantozziLambertLiembergerSanPaolo7AprIWA-
2010K.pdf
At least one large French Water Company with international activities has been
calculating ILI for several years and using it as one of their key performance indicators,
and has provided 4 ILIs for French cities to the European ILI data set for HICs.
Germany The IWA Water Balance and several IWA concepts were adopted in DVGW
W392 Guidelines in 2003, but DVGW preferred to continue to use the traditional German
performance indicator of specific loss (m3/km/hr) as a performance indicator, whilst
continuing to strongly recommend against the use of %s. However, after re-evaluating the
advantages of the ILI, the draft 2013 update of W392 now recommends adopting the ILI
as the definitive PI for Real Losses, with specific loss as a subsidiary measure, for the
6000 Water Utilities in Germany.
ILI values interpreted visually and very approximately from a graph in Tennhardt (2012)
suggests that around 70% of a sample of 35 calculated German ILIs are within Band A,
with most being grouped around an ILI of 1.0 +/- 0.5. The remainder are mostly in the
range 3.5 to 4.5. Water balance data grouped within 4 of the 16 German Bundesländer
(States) has been derived from Consultants reports and assembled by Koelbl and
Lambert, and contributed to the 2014 data set on an anonymous basis. The grouped ILIs
range from 0.7 to 1.1.
In Italy, the IWA methodology and the ILI are now well known through training courses,
workshops and Conferences initiated by Dr Marco Fantozzi, and based mainly at Genova,
Ferrara, and Italian Universities; these have taken place on a regular basis since around
2004. A data set of 67 partly validated Italian ILIs in Piedmont, Emilia Romagna and
Tuscany, ranging from 1.5 to 16 (median 5.0) were collated by Marco Fantozzi in 2008.
Enia Utility (now part of IREN Utility) is one of a number of Utilities that has been using
IWA practical approaches for up to 9 years, and has contributed good quality ILI data from
10 towns (average ILI 2.8) to the 2014 ILI data set; an ILI from 1 anonymous Italian city is
also included. The regulator in Emilia Romagna now requires Utilities in that region to
calculate ILI, and there is growing interest at National level to consider moving from the
traditional %s and losses per km as performance indicators .
References: Calza F and Fantozzi M. Pressure management: the experience of ENIA Reggio
Emilia, WATER EFFICIENCY CONFERENCE, Ferrara, Italy May 2010
Calza F and Fantozzi M: Experiences in district metered areas and pressure control in ENIA, IWA
WATERLOSSEUROPE CONFERENCE, Ferrara, Italy May 2012
- 28 -
Malta Water Services Commission adopted the IWA Water balance and began to use the
ILI in 2001 as its key performance indicator for Real Losses. In 2007 the Malta Regulator
(the Malta Resources Authority) formally recognised the ILI by setting WSC’s performance
targets in terms of ILI. Malta has contributed ILIs to all three IWA data sets – 1999, 2005
and 2014 – and the overall ILI for Malta WSC has fallen from 10 in the mid-1990s to 2.0 in
2012 (Galea St John, 2013). Because all properties have customer storage tanks, ILI
calculations are based on night flows using ‘Snapshot’ ILI. The WSC has contributed 2
ILIs to the 2014 data set, for the islands of Malta and Gozo.
th th
Galea St John, S (2013). Awareness of Leakage & NRW in Malta. Global Leakage Summit, 12 /13 March
2013, London,
Portugal The single ILI for Portugal in the European data set was a late entry, and a more
informative overview of progress in implementing the IWA Best Practice approach will be
provided in an update in later in 2014. Most systems operate with direct pressure, Porto is
one of a limited number of Portuguese Utilities where customers have roof storage tanks.
Serbia Progress in the introduction and implementation of the IWA Practical Approach is
occurring mainly through inter-municipal cooperation developed and supported by the
German development agency GIZ (previously GTZ). A 2010/2011 implementation project
provided rising awareness of the IWA approach, and free Water balance and PI software
was provided in Serbian language. Activities included promotional presentations,
workshops, water network audits, training of employees from utilities, on-field education,
webinars and visits to good example case studies (even between different countries).
Altogether 8 utilities were directly involved with participating personnel from another 20
utilities, with an IWA water loss expert as main consultant and trainer.
IPM is a new association of professionals from the Serbian water sector in Serbia, formed
to help Water Utilities’ future cooperation, to promote efforts and achievements with
increase public awareness, and to seek political support and financial assistance from
different donation programs and associations. Associated initiatives include the Water
Loss Working Group of the Water Operators' Partnerships for South East Europe (WOP-
SEE), and publications promoting IWA methodology and translations of Water 21 articles.
Jurica Kovac has collated ILI data from 7 Utilities as Serbia’s contribution to the 2014
LAMIC European data set.
Spain There are around 2800 Water Utilities in Spain, and many municipalities operate
their own systems. Most of the properties receive direct pressure from the mains, but in
some areas of the East coast of Spain there are cities where the properties have roof
tanks. The benefits of pressure management are well known in technical forums, and all
large cities have pressure management schemes, some with over 100 pressure managed
zones. Economic benefits including some reductions in burst frequencies are being
obtained, but there is little published data and reports.
The Spanish Water Supply Association publishes a book of water supply statistics, with
figures of average NRW%, so this remains the most commonly used performance
indicator. Other figures, performance indicators and practices are generally published or
delivered. Some Companies calculate an ILI but do not publish it, and one of these
Companies has provided an anonymous ILI for the European data set.
- 29 -
Switzerland The single ILI for Switzerland in the European data set was a late entry, and
a more informative overview of progress in implementing the IWA Best Practice approach
will be provided in an update in later in 2014.
The Swiss guideline SVGW W4 “Water distribution”, published in 2013, consists of 5 parts
(Part 1: General, Part 2: Planning and Design, Part 3: Construction and Control, Part 4:
Operation and Maintenance, Part 5: a collection of loose sheets of examples from
practice, check lists etc.). Water Losses are addressed in Part 4. Prior to 2013, calculation
and assessment of water losses were not described in any other guideline.
In the SVGW W4 guideline it is recommended to use the IWA Water Balance. For
performance indicators, the guideline follows the DVGW W 392 (2003) guideline approach
with a classification of qVR (m³/km/h) for rural, urban und large urban networks, with class
limits based on system input rate. However, the Guideline does not describe details of
water balance and performance indicator calculations. Some utilities have started
calculating the ILI for the purpose of international comparisons, but the use of
percentages and qVR is still widespread.
United Kingdom There are only around 25 Water Utilities in the UK, half of which are
very large organisations. Leakage calculation based on Water Balances, Night Flows and
associated data for England and Wales are independently audited. Until 2009/10, three
separate figures for leakage in each Company were published annually by OFWAT, the
Economic Regulator, each in 3 sets of units: Ml/day, litres per property per day and m3/km
of mains per day.:
Distribution losses, up to the property line
Total supply pipe losses, on privately owned pipes after the property line
Total leakage, the sum of Distribution Losses and Total Supply Pipe Losses
OFWAT clearly stated in 1996 that leakage as a % of Water into Supply can be
misleading, as a reduction of volume into supply, following successful water efficiency
measures by a Utility’s customers, would make its leakage performance appear to
worsen.
In 1996, OFWAT began to set England/Wales Companies targets for Total Leakage,
based on calculations of economic intervention policies for active leakage control and
Economic Leakage Levels (ELL) carried out by the companies. These were broadly
achieved by 2000. However, it is now recognised that the ELL calculations did not take
enough account of the several influences of pressure management (which is widely
practised in the UK). Prediction methods for the Sustainable ELL (SELL) now being used
by the Companies include allowances for costs of social disruption and cost of carbon, but
there are concerns that pressure management benefits and meter location issues
influencing supply pipe leakage are still not receiving sufficient weight in most SELL
calculations.
OFWAT ceased to publish detailed water balance and associated data in 2010, and
introduced a risk-based approach to regulations. As part of this new way of regulation they
introduced a number of key indicators which includes Leakage. From 2011-12, the only
Company leakage figure published on the OFWAT Website is ‘Total Leakage’ in Ml/d,
usually derived from both ‘top-down’ water balance and ‘bottom-up’ night flows. In the
absence of other traditional UK leakage performance indicators previously published by
OFWAT (litres/property/day, m3/km mains/day), the media fill the gap by publishing their
own diverse interpretations of the Ml/d figures, which tends to confuse public perceptions
of leakage management performance.’
- 30 -
The Environment Agency also collects leakage data and requires the companies to
develop Water Resource Management Plans, which contain details of current leakage and
leakage targets within the context of the supply demand balance.
Leakage targets in Scotland are agreed between Scottish Water and WICS (the Water
Industry Commission for Scotland). Water Resource Management Plans are also
published in Northern Ireland and Scotland.
Since 2003 a number of UK Water Suppliers (including Northern Ireland and Scotland)
contribute leakage performance data to a Water UK Leakage Network group, but this
unaudited data is not published or available to researchers or the media. All members of
the Leakage Network can now calculate their ILIs using free software provided by WLSG
members, and nine members have contributed ILIs to the European data set on an
anonymous basis.
- 31 -