Three long draft reports have been issued in the past few months on Water Loss KPIs – two in the UK and one by the Inner Group of the IWA Water Loss Specialist Group. These are

• Managing Background Leakage: Stage A Report Final February 2023 Invenio, 5 UK Water Companies and University of Sheffield, UK
• Understanding Factors Contributing and Affecting DMA Leakage by UK Water Industry Research (UKWIR Report ISBN 978-1-838-045-2, 2026) 2026: being delivered in phases in December 2026 with full findings disseminated in 2027
• Guidance Notes: Best Practices and Challenges in the Use of Water Loss Key Performance Indicators (Draft 4.1, 18th September, authored by the Trow, Wyatt, Pearson and Koelbl, of the IWA Water Loss Specialist Group (154 pages). Comments are required by 31st October (7 days from now) but I found too many factual errors in this document to meet this tight timescale.

The ‘Gold standard’ leakage KPI for comparing leakage management technical performance in large distribution systems is the Infrastructure Leakage Index ILI ,  developed by the 1^st IWA Water Loss Task Force (1995-99) for large systems with more than 5000 service connections at mid-range pressures.  ILI is the ratio of Current Annual Real Losses CARL divided by Unavoidable Annual Real Losses UARL. UARL is the sum of three major pressure-dependent components of real losses (Reported Bursts, Unreported Bursts and Background (undetectable) Leakage. UARL includes basic Company management policies with clearly stated objectives to limit the durations of reported and unreported bursts. Many large Water Utilities internationally have now achieved ILIs close to 1 (CARL = UARL) at their current operating pressures. ILIs are non-dimensional so can be compared anywhere worldwide.

Let me declare an interest here. I had the honour and considerable pleasure of chairing, on a pro bono basis, the 1st IWA Task Force which developed UARL1999 and ILI, with a supportive Committee of senior technical experts appointed (not self-selected) by the American, French, German and Japanese Water Associations. I also met many of the best leakage specialists worldwide in many different countries where we shared knowledge on a co-operative (not competitive) basis. UARL1999 and ILI were initially applied to 27 large systems (including several National and State Capitals) in 19 countries before any publication, and some were found to be losing significant multiples of their UARL1999 volumes in real losses. Some countries and Utilities immediately accepted the challenges and have since achieved major sustained reductions of both leakage and ILI; others have preferred to try to ignore or downplay the message ,and still do.

An ISO standard records progress since 2000 in defining boundary conditions for the application of UARL to cover smaller systems (including DMAs) outside 45 to 60 metres pressure with calculated ILIs less than 1. For these systems, UARL1999  is customised using System Correction Factor SCFs, which are based on further applied research linking leak flow rate sensitivity to pipe materials using FAVAD (Fixed and Variable Area Discharges concepts), use of median annual rather than average annual burst flow rates, and influences of pressure on burst frequencies.

These local factors work together in different combinations which increase (or decrease) achievable leakage . Sometimes they reinforce each other and sometimes cancel each other out,  which is why minimum leakage achieved in small systems and DMAs is so variable. Customised  System Correction Factors (SCFs) have already been applied to several thousand large and small systems worldwide – all of Quebec Province, Malta, Sydney Water and another large City Utility in Australia, from the smallest DMAs to the whole Utility . Also to DMAs in the UK, USA and other countries. Yet UARL with SCF gets no mention in the IWA WLSG Report.

The UK Position on ILI

Maybe it is unsurprising that these 3 latest Reports which push CRLI and denigrate UARL and ILI originated in the UK or have strong UK representation. UK has never formally calculated or published ILIs; the economic regulator and the Leakage group of UK Utilities has simply refused to formally or informally acknowledge this KPI for 26 years.  However I do have some ILI values provided to me by several of the more outward looking Utilities in England and Wales who contributed to the IWA 1999 data set, and the 2014 European ILIs data set in the EC ‘Good Practices Study’ in 2016. These mostly show ILIs in Band A (ILI greater than 1 but less than 2.0 ).  However others Utilities in UK remain  reluctant to provide such data, even when free software is provided.

Historically most households in UK were unmetered, and relied heavily on interpretation of DMA data, but this is now moving towards a mixture of external meters (at the street boundary) and internal metering within households. Assessment and publication of of leakage in m3/km of mains per day and litres/property per day (not per service connection as recommended by IWA WLSG) is  based both on night flow analysis of DMAs, and annual Water Balance using estimates.   My understanding is that if the difference between these two figures is less than 25% of Water Supplied, the annual leakage is assessed as somewhere in between them. This is not transparent process to outsiders.

Stuart Trow and I were active within the Water Supply group of the Institution of Water and Environmental Management in UK some 10 years ago, and (together with several IWA WLSG colleagues) in the 2016 ‘EC ‘Good Practices’ Report which recommended that ILI should be calculated in UK, but both these recommendations have been ignored, even though UK is a member of EUREAU.

UK did make considerable contributions to international leakage management concepts from 1980 onwards   regarding pressure management  to reduce leak flow rates (Report STC26 Leakage Control Initiative, and John May’s Fixed and Variable Area Discharges (FAVAD) concept for pressure:leakage relationships.  Also Component Analysis of Annual Real Losses and Night Flow in 1995 which I and others developed for the UK National Leakage Control Initiative 1991-95.  However, UK has since 1995 used higher values for  background leakage than the IWA ‘low’ standards in UARL used by other High Income Countries  (World Bank Rankings).

However, on the basis of night flow tests on service connections in 25 small DMAs in the first Phase of the 2026 UKWIR Report analysed using CRLI, the UK conclusion now is that background leakage on service connections is likely to be almost zero and component analysis of real losses is an outdated concept even for large systems.  I feel sure that others internationally would wish to see some robust ILI calculations for UK before endorsing that view!

CRLI: A Work in Progress

Despite the relentless promotion of CRLI and denigration of UARL and ILI using selective ‘facts’ in these three Reports, I quickly recognised that the first calculations of CRLI values made by its originator (Wyatt) were based on square root curve fits to the UARL equations for litres/connection/day and m3/km mains/day vs density of connections at 50 metres pressure when customer meters are located at the edge of the street.  I had provided him with that base information to assist his early research several years ago. He has acknowledged this fact privately (but sadly still never publicly in presentations or articles). When promoting CRLI  it seems it was just too tempting to relabel the UARL equation for a specific situation – meters at the edge of the street – and promote it as a new concept for  wider international applications which it does not cater for. But why then disparage UARL and ILI?   CRLI should be subject  to the same robust critical examination that UARL and ILI have experienced over 26 years, if it wishes to survive on its own merits.  

CRLI can be described at this time as a work in progress, not  a finished product, despite the glowing endorsements in the three documents. When the logic of CRLI is questioned and found to be wanting,  the goalposts tend to be moved and that particular claim is quietly dropped in the next report and description. Here are some examples:

• CRLI was initially promoted as a geometric mean using an over complicated formula (square root of litres/connection/day  times  m3/km/day), which (it was said) did not use or need the density of service connections per km of mains, or average pressure.  However this equation actually ‘hides’ the presence of density of connections
  • Simpler forms of the CRLI equation easier for practitioners to understand are
    • CRLI = 0.0316 x litres/connection/day x SQRT (Density of Connections)
    • CRLI = 31.6 x m3/km mains/day divided by SQRT (Density of Connections)
• CRLI does not meet all the requirements of a geometric mean
  • The two values litres/conn/day and m3/km mains /day should not be less than or close to 1 and should have similar magnitude; however in CRLI they often do not. particularly at high or low connection densities
  • The geometric mean cannot be negative or zero for real-valued data (e.g litres/service connection/day and m3/km mains/day)
    • If the geometric mean principle is extended to include 3 components of leakage (for example, leakage from private supply pipes up to customer meter) CRLI is the cube root of product of three values. If one of these values is sometimes zero, then CRLI will be zero. So CRLI is only valid where customer meters are always at the edge of the street.
    • In the UKWIR Report (page 38) when UK Companies reported negative leakage on the service connections (night flow less than estimated night use) the CRLI equation was subjected to ‘small modification’ –  division by a modulus, which changed any negative CRLIs to zero. While this allowed that DMA data to be included in the DMA data set when assessing background leakage, it ignores that the criteria for using geometric means (CRLI not equal to zero) will have been breached. ; I pointed out that the valid option would be to delete the ‘small modification’ when calculating CRLI, and delete the  DMA from the background leakage analysis.

The IWA Guidance Notes (Page 36) acknowledges (with respect to both the top down water balance and the Night leakage calculations) that negative leakage values can result. This is followed by the extraordinary statement that ‘because two negative values are multiplied the CRLI will be positive’. Thankfully the recommended approach on page 36 of the draft IWA Guidance Note has now changed, it  is now to screen out negative real loss values before calculating the CRLI. The ‘small modification ‘ of the modulus in the CRLI equation now also needs to be removed in the two UK Reports.

What confidence can readers of the Reports have if this type of deliberate data manipulation has been permitted, allowing CRLI to be promoted as the chosen scaling factor in the UKWIR Report, with the recommendation that ‘CRLI has been shown to be a very effective normalisation at DMA leakage and it is recommended this is taken forward in future projects’ . Particularly when the other options of litres/connection/day and m3/km mains per day were passed over with scant superficial consideration, and UARL or UARL with SCF (for annual losses from Water Balance) were never mentioned.

Another source of concern is that CRLI is being introduced in the new first (and probably subsequent) ‘UK Managing Leakage’ Reports in a country which has always been dismissive and even antagonistic to IWA international concepts. This is then used to promote CRLI with concepts unfamiliar to IWA practitioners such as Bayesian Statistics (which I am reasonably aware of through my years in Hydrology and Water Resources ) in the UKWIR Report.

A Wake-up Call: Evolution, Not Revolution

And now we have the IWA Guidance Notes: Best Practices and Challenges in the Use of Water Loss Key Performance Indicators (Draft 4.1, 18th September), authored by Trow, Wyatt, Pearson and Koelbl, of the Inner Group of the IWA Water Loss Specialist Group. This promotes CRLI internationally even for leakage calculated from Water Balances without producing any evidence other than from the UK reports , whilst denigrating UARL, ILI and UARL with SCF at every opportunity. Comments are required by 31st October only 8 days from now, giving IWA members insufficient time (43 days) to comment on a 154 page report with many internationally untested ideas.

The 1^st IWA Task Force had 5 years to lay strong foundations for understanding large systems, then another 20 years of applied research by  many WLSG members and colleagues such as John May, who changed the understanding of why flow rates from some kind of leaks varied with pressure by up to three times the previous international assumptions (square root)  that had lasted for 200 years. Or the fact that we can now predict the influence of changes in burst frequency from changes in system pressure, and predict the savings from extension of asset life, which most thought was impossibly difficult only 20 years ago; this transforms calculations of economic leakage management investment.

In the 40 years I have been researching, unpicking and then bundling together (for analysis and predictions) the physical and hydraulic principles that contribute to the  leakage from distribution systems,  and then combining them in practical ways for leakage practitioners, I have been much influenced and impressed by the methods of an American called Richard Feynman (now deceased) who has helped many people to understand complex problems. One of his many popular quotes is that ‘the easiest person to fool is yourself’.  Much of his work,  which he  always explains in ways which an interested  young person with a basic education could relate to, can be can  be found on YouTube.

You will probably realise by now that the ‘Wake-up call’ I’m referring to in the title to this personal piece is that IWA WLSG members and others who truly value leakage management as a major contribution to water conservation in this century of huge challenges need to Wake Up to the fact that they are being rushed into taking hasty decisions on biased information largely from a country – sadly my own birthplace – which has never accepted proven IWA WLSG concepts. Evolution, building slowly on the foundation of the proven methods which so many of my IWA colleagues have pioneered, and which have saved so much water over the last 25 years, is surely the way forward – not a CRLI revolution. The ground-breaking work of the IWA Water Loss Task Force and Specialist Group must not be compromised by hasty decisions. 

To all readers of this personal view – thanks for staying with me to this point. I have seen enough in my long career in leakage management to recognise when I am being ‘steered’ with questionable selective ‘facts’ and ‘woolly arguments’  which do not stand up to critical examination. So question everything, including motives.  Don’t take something for granted just because there is a report with  academic content and titles (Geometric Means, Bayesian Statistics, …..) which you may find intimidating rather than enlightening. Don’t hesitate to dig deep and ask awkward questions. There is no such thing as a stupid question, as often the answers they produce are found to be unconvincing.

Allan Lambert  23rd October 2026

Please note previous opinion piece on CRLI here