Leakage-driven mains renewal

Dennis Grimshaw, technical development director of RPS Water, reviews the current leakage position and looks at mains renewal as a preventative measure

Evidence from water undertakings in England and Wales suggests that the marginal costs of active leakage control are increasing to uneconomic levels in some parts of their distribution networks as a result of asset deterioration. This has prompted a move away from purely water quality driven mains rehabilitation towards a greater focus in AMP4 on leakage savings through mains renewal.

This article reviews the current leakage position in England and Wales, assesses the impacts of the proposed AMP4 mains renewal rates on leakage, and seeks to examine the network conditions under which mains renewal would be a cost-effective alternative to find and fix; an analysis framework based on the natural rate of rise in leakage (NRR) is proposed.

The information gaps which need to be bridged to ensure effective targeting and the uncertainties associated with mains renewal as a leakage control policy are also summarised.

Since the mid 1990s, leakage targets in England and Wales (based on the economic level of leakage, ELL) have been successfully achieved (for most companies) by applying a policy of proactive find and fix (active leak detection and repair) as the most cost-efficient approach.

Find and fix has also been augmented by pressure management in many companies as a means of producing stable and optimal pressures to meet regulatory standards of service, reduced background levels of leakage and reduced burst frequencies.

Leakage levels in England and Wales (Figure 1) have been reduced by more than 1,500Ml/d (almost 30%) over the past ten years. However, despite the successes for most companies (because of the poor serviceability of the network infrastructure) there are claims that maintenance of the current low levels, and any further significant reductions, can only be achieved by promoting asset renewal as a control policy alongside find and fix and pressure management.

Characteristics of leakage

The total leakage reported annually by water undertakings in England and Wales is made up of estimated losses from trunk mains and service reservoirs, the distribution network and losses from customer supply pipes. Losses from the distribution network can be attributed to leaks from the distribution mains themselves, the communication pipes (from the main to the property boundary) and the various fittings such as valves, hydrants and stop taps.

In the context of mains renewals, it is common practice to rehabilitate mains and associated communication pipes, both of which are in the ownership of the water undertaking. Supply pipes are currently within the ownership of customers and there is currently much debate in the industry as to whether this is appropriate given that customer-side leakage represents a significant untapped resource.

A significant proportion of distribution system leaks is visible and is reported by members of the general public or others. This typically accounts for about 60% of all leaks repaired by a water company. The remainder are detected by company staff and their contractors using combinations of sounding, acoustic logging and correlation techniques. There is a continuing search in the industry for improvements in detection technology and practice to improve efficiency and reduce the lower limits of detection (the so-called background level of leakage).

The level of detection and repair activity required to maintain a given leakage level is dictated by the natural rate of rise (NRR) in leakage resulting from the breakout of new leaks and the increase in the flow rate of existing leaks. Background leakage and NRR are two critical parameters in leakage economics and both are dependent to a large degree on asset condition.

Drivers for mains renewals

The principal drivers for most mains rehabilitation or renewal programmes will include:

  • The need to overcome current and projected future deficiencies in asset performance and service delivered to customers measured against accepted standards
  • The need to extend the serviceable life of an existing group of assets

These are achieved by employing an appropriate combination of rehabilitation techniques matched to the nature of the problems to be solved and the physical and operating characteristics of the distribution network.

Selection and prioritisation of target sections of a mains network to be rehabilitated is governed by the current and projected trend in the values of a number of serviceability indicators, for example low pressure, interruptions to supply, iron discolouration. The challenge is to restore the assets to a position where the service delivered to customers is acceptable (and affordable) at minimum whole-life cost.

The UKWIR Capital Maintenance Planning Framework (CMP)2 guidelines represents industry best practice for developing investment plans for asset capital maintenance of which mains renewals can be considered a specific element. This approach has been used by water companies for their PR04 business plan submissions for the current Asset Management Period 2005 to 2010 (AMP4). The CMP Framework is a common-sense risk-based approach that relies on the development of relationships between asset performance and service to customers as a basis for selecting investment levels to bridge serviceability gaps.

The most recent UK water mains rehabilitation programmes have been driven by water quality considerations (under Section 19 Undertakings). However, this large-scale programme is now coming to an end for most undertakings in England and Wales and other drivers are now increasing in significance. In particular, some water undertakings have justified their PR04 submissions for mains renewals on the grounds that leakage control in some sections of their distribution networks has become uneconomic using traditional find and fix techniques.

Published data on water company levels of mains renewals planned during AMP4 (excluding mains relining) gives the following totals:

Water and sewerage companies – 3,036km per year – 1.09% pa

Water only companies – 380km per year – 0.70% pa

The current distribution of mains infrastructure “year laid” for water and sewerage companies in England and Wales is illustrated in Figure 23. This has been averaged (excluding Thames Water) to provide the typical water company infrastructure age profile in Figure 3. This illustrates that (excluding Thames Water), on average about 45% of water companies’ mains infrastructure is more than 50 years old and about 5% of mains are more than 100 years old. The overall average age is just under 45 years.

The effect of different annual rates of renewal on the average age of mains is shown in Figure 4, assuming that the oldest mains are targeted first. For the average age profile in England and Wales (excluding Thames) it may be seen that the long run rate of renewal required to maintain the current average mains age is just over 1% a year. This corresponds to the proposed AMP4 levels of renewal indicated for water and sewerage companies (provided that this is sustained) but is a higher rate than that proposed by water-only companies.

Impacts on leakage


Conventional wisdom suggests that in average circumstances it is unlikely that mains renewal would be economically justified as a blanket alternative to find and fix on the basis of leakage savings alone. However, a commonly held view is that there are hot spots or individual lengths of main within a network where mains renewal would be an economic alternative. These hot spots are likely to be characterised by high background leakage levels and high NRR since both these parameters are inherently dependent, at least in part, on the condition of the asset.

In a typical DMA where infrastructure renewal is planned, there is an expectation that leakage benefits will be realised through:

a) Volumetric savings

  • Removal of a major proportion of the current leakage in the renewed lengths
  • A significant reduction in the total natural rate of rise in leakage

b) Cost savings

  • Detection costs (transition and maintenance)
  • Repair costs

The degree to which the above benefits will be realised and sustained will depend on the rehabilitation techniques employed and which assets are targeted. Mains renewal or rehabilitation by definition is largely targeted at distribution mains (and usually the associated communication pipes) and anecdotal evidence suggests that the majority of leakage problems are associated with the stock of very old cast-iron mains, AC mains and perhaps older uPVC mains.

However, customer supply pipe leakage is often underestimated as a percentage of total leakage and therefore a sizeable proportion of potential leakage savings may not be realised, without a parallel programme of supply pipe replacement. This can have a major influence on the selection of water main rehabilitation technique and hence unit cost and productivity. However, if customer supply pipes are clearly shown to be a significant cause of leakage in DMAs targeted for rehabilitation, then replacement of these should be considered in parallel.

Against these potential benefits, there needs to be an awareness of the possible impacts on the areas of the network not selected for renewal and the wider area as a whole. Case studies have shown for example that mains renewal may increase leakage and possibly mains burst frequencies as a result of:

  • Increases in static pressures in the network
  • Increases in the degree of disruption and transient pressures in the network during the works themselves

Both of these may generate a need for increased detection and repair activity or pressure management. In addition, leak detection activity on customer supply pipes in the renewed areas will still be required and DMA leakage monitoring activity is unlikely to be reduced. There are also potential impacts on water quality and social disruption during the construction works. It is therefore essential that, even when the driver is just leakage, mains renewal is planned not in isolation but is appraised as part of an integrated leakage solution including find and fix, pressure management and possible re-zoning options.

Provided that these impacts are managed, there are other potential long-run impacts on the zone network as a whole whereby zone average background leakage and NRR are likely to be lowered and as a result may promote a lower economic level of leakage.

Current leakage levels in England and Wales

With a few notable exceptions, most UK water companies are now operating at or in some cases below their economic levels of leakage, and the challenge now is for most companies to maintain these levels against a background of deteriorating infrastructure serviceability and pressure by some regulators for further reductions in leakage.

There appears to be agreement, however, that economic levels of leakage will continue to be the basis of target setting for the foreseeable future. The gaps between current (2004/05) published levels of total leakage and the targets set by Ofwat for the end of 2009/101 are shown in Table 1. This shows that in volumetric terms a reduction totalling 272Ml/d is required over the AMP4 period. The greatest burden has fallen on the water and sewerage companies and of these almost two-thirds of the required saving is targeted at Thames.

Comparisons between the leakage performances of different companies is always fraught with difficulty because of the different physical and operating characteristics of the distribution networks, marginal cost of water and in particular, network density.

For this reason, company leakage figures are usually reported relative to numbers of properties supplied and also length of mains. In the context of mains renewal policies it is useful to consider leakage measured in units of m3/km/d and Figure 4 shows the 2004/05 reported leakage figures plotted by inverse network density (mains length per connection, m).

Figure 4 illustrates that:

a) For a given network density, W&SC and WOC leakage figures are comparable

b) There is a not unexpected relationship between leakage and network density – as the mains length per connection increases in the more rural networks, leakage in m3/km/d decreases as the contribution of leakage from supply and communication pipes reduces

The 2009/10 leakage targets can be plotted in the same way (Figure 6). Notwithstanding other company differences, such as infrastructure age and condition, the targets, even for Thames Water, appear to follow a logical trend with network density. This is further reinforced by ELL studies undertaken by the author in the Republic of Ireland where county distribution systems are generally much more rural in character. ELLs are plotted on Figure 6 for three counties in the republic where mains length per connection is 30-50m.

The leakage values used in Figures 5 and 6 represent total leakage – the sum of leakage from customer supply pipes and leakage from the company infrastructure, largely distribution mains and communication pipes. The estimated proportion of total leakage represented by distribution mains and communication pipes is shown in Figure 8 (by deducting published levels of customer side leakage) and again illustrates the expected increasing proportion of leakage from these assets in the more rural networks. For the range of average company network densities in England and Wales, this would vary between 60% and 70% for highly urbanised networks to nearer 80% in the more rural networks. Unlike find and fix, mains renewal can be expected to reduce the background leakage in the renewed assets.

The current total leakage in England and Wales is equivalent to 11m3/km/d. Assuming that 28% of this figure is customer supply pipe leakage, and assuming that a further proportion of say 10% is trunk main and service reservoir leakage, the average mains and communications pipe leakage could reasonably be assumed to be about 6.5m3/km/d.

So, provided that the side-effects of mains renewal activity is properly managed, it would not be unreasonable to assume that at least this level of leakage could be saved, 6.5 x 3,416km/1000 = 22.2Ml/d per year. Also, given that mains renewal would be targeted at the leakiest assets, rates of possibly two or three times this might be expected giving an overall total similar to the target leakage reduction of 270Ml/d required of companies over the AMP4 period. Clearly further savings could also be realised if a customer supply pipe replacement programme was undertaken in parallel.

Natural rates of rise in leakage

The total natural rate of rise in leakage (NRRt) is usually defined as the hypothetical annual increase in leakage that would occur if no leak repairs were carried out. In reality, even if the leakage control policy is one of reaction only to reported or visible leaks, then at least these would be repaired. And so the detected proportion (NRRd) becomes the additional leakage that would need to be overcome each year for leakage just to stand still.

RPS Water has recently completed a UKWIR study into NRR, which has included the development of a best practice guideline for calculating NRRt and NRRd from company data. At the time of writing, the UKWIR report on this work is awaiting publication.

There is currently no published relationship between NRR and distribution network characteristics. And the recent UKWIR study, because of time and budget constraints, did not attempt any development of this relationship. However, there is a wealth of evidence to indicate that NRR is a function of the following physical and operating characteristics:

NRRt = ’ {Lm, Nc, Ns, P, If, Df}


Lm – length of mains

Nc – number of communication pipes

Ns – number of customer supply pipes

P – pressure

I – “infrastructure serviceability”

D – “network disturbance”

The If factor has been rebadged “serviceability” rather than

“condition” as a means of distinguishing between mains which may be considered for example to be in “poor” condition but are still serviceable and are not disposed to leak. The converse may of course also be true.

The Df factor is intended to reflect the anecdotal evidence that high levels of regular disturbance within a network, for example through operational maintenance activities or large diurnal pressure variations, will in themselves result in increased breakout rates.

The actual functional form of the relationship has yet to be developed within the industry but a hypothesis might be that it is similar to the Managing Leakage4 equation for background leakage. A functional form similar to the following is therefore proposed:

NRRt (l/hr per year) = PCF . [(m . Ifm . Lm ) + (c . Ifc . Nc)] . Df


PCF – Pressure Correction Factor

m – the average contribution of mains (and mains fittings) leaks (l/km/hr) to NRR

Ifm – Mains infrastructure serviceability factor

Lm – Length of mains (km)

c – the average contribution of leaks from comm. pipes and service connections and fittings, (l/connection/hr) to NRR

Ifc – Communication pipe and supply pipe infrastructure serviceability factor

Nc – Number of connections

Df – Network disturbance factor

A small number of UK water companies provided data for the UKWIR study and the typical calculated DMA level NRRt values, pooled and averaged over a range of network density intervals, have been used to give indicative values for the m and c values in this equation. These are:

m = 90 l/km/h per annum

c = 8.8 l/connection/h per annum

No attempt has been made at this stage to provide a range of values for If and Df and further work is needed to fully evaluate this model. However, as an illustration, the variation in NRRt with changing If and Df can be expected to be similar to Figure 8.

Figure 9 illustrates the NRRt and resulting proportion of NRRt (converted to m3/km/d) contributed by the mains, comms pipes and associated fittings for a range of network densities. For the typical range of England and Wales, water company densities, average company level NRRt values would be expected to lie between 21 and 12m3/km/d of which the mains and comms pipe values would be expected to lie between 12 and 7m3/km/d. Clearly there is likely to be a wide range of values around these figures at local zone and DMA level dependent on pressure, asset serviceability and network disruption effects.

Provided that the serviceability and disturbance coefficients can be quantified for typical networks, the NRRt function may be used as a means of assessing the annual savings attributable to the renewal of specific lengths of main or hot spots within a network. NRR will vary spatially across any given DMA, depending on the age and serviceability of the infrastructure. For example, if we take leak breakout rates as a surrogate for serviceability, there is some published data on the relationship between infrastructure age and mains failure rates5 which may be used to provide an illustration of the spatial distribution of NRRt across a network. Ideally, a full spatial analysis of leak repairs and mains material would be used for this but using this failure data would reveal a distribution of NRRt with infrastructure age similar to that shown in Figure 10.

The age profile of the mains infrastructure for water companies in England and Wales has been illustrated in Figure 3 and combining this with Figure 10 would result in a distribution of NRRt with the variation of asset age across a typical water company network as shown in Figure 11. This should be regarded as a typical profile and may be determined for any given network from a spatial analysis of leak repair and mains material data.

Temporal change in NRR

A distribution of the type illustrated in Figure 11 for a given network can be used to model the effects of different renewal rates on the NRR and hence change in leakage level over time. For example if we assume no infrastructure renewal, Figure 12 illustrates the change in the average network NRRt that could occur as a resulting of the changing age profile and assumed corresponding decrease in asset serviceability.

In the short-run, over say ten years, this would add 2.7Ml/d per 1,000km to total leakage, aside from additional leakage due to increases in background leakage.

For a range of renewal rates (again assuming the typical infrastructure age profile shown in Figure 3), Figure 13 shows the shift in the average network NRRt. The current average NRRt for this age distribution is 18.7m3/km/d and from Figure 13 it may be seen that in the short term (ten years) a renewal rate of between 0.4% and 0.5% pa would be required to maintain the current network NRR. However, in the medium to long term, a renewal rate of between 0.7% and 1.0% would be required, all other things being equal and assuming that supply pipe leakage is controlled by other means, for example find and fix.

Given the proposed AMP4 rates of renewals of an average of 1.1%, the average NRRt can be expected to reduce from 18.7 to between 15 and 16m3/km/d per year.


The above calculations suggest that the AMP4 annual renewal rate of just under 1.1% for W&SCs would (if properly targeted) be at least sufficient to hold natural rates of increase in leakage steady (at least in the short term) and may result in a net reduction in the overall network average.

This is again heavily dependent on the potential side-effects outlined earlier being properly appraised and managed. However, the question may be asked whether this is the most cost-effective policy for controlling leakage.

It is well known that the cost of reducing leakage through active leakage control (detection and repair) increases as the level of leakage decreases – it becomes harder and more expensive to achieve very low levels of leakage. A typical relationship for leak detection activity is illustrated in Figure 14.

A similar relationship can be developed for repair activity but in previous studies undertaken by the author, marginal repair costs have been found to be much less dependent on leakage level.

In ELL studies, and in particular the modelling of the effects of mains renewals, it is useful to express this relationship in total annual cost terms. The MELT equation6 (Annual Cost = c . [(L1+NRR-BL) d – (L2-BL) d] £ per km per yr. Where: L1 and L2 are start and end levels of leakage (m3/km/d) c and d are coefficients) provides an ideal functional form for this.

Each of the marginal cost curves is asymptotic to the background level of leakage (BL), and as this increases with deteriorating asset condition it can be expected that the marginal cost to maintain a given level of leakage will increase, as is currently being observed by some water companies. The question is at what point does this become uneconomic and mains renewal become an attractive alternative?

Comparison of long-run marginal costs

To try and illustrate this, the long-run marginal costs (LRMC) of active leakage control (detection and repair) and mains renewal have been compared for a range of NRR and background leakage levels.

The LRMC of active leakage control is estimated to be in the range 40p to 60p/m3 for leakage levels in the region of the ELL. This assumes that about 60% of leaks in a distribution network are visible and are reported by members of the public and others. The calculation has also assumed that NRR will deteriorate with infrastructure age according to the profile shown in Figure 12. The environmental and social costs of each option have not been included in this example.

The LRMC for mains renewal (in the context of leakage savings only) is dependent on:

a) The current level of mains and comm pipe leakage that would be saved (less any deterioration in the renewed mains)

b) The increasing natural rate of rise in leakage that would have occurred in the mains and communication pipes, year on year had they not been renewed

c) The cost of renewal

Given a typical mains age profile such as that shown in Figure 3 and the corresponding NRRt levels (as in Figure 11), a range of LRMC’s for mains renewal has been calculated assuming that:

  • About 72% of NRRt is saved – only the mains and communication pipe components
  • The cost of renewal is taken as the present value (PV) cost of advancement from the year in which the main would have been replaced anyway – at the end of its serviceable life (assumed in this case to be 140 years), and now
  • The average unit cost of renewal is £150/m

The resulting LRMC values for a range of advancements (and corresponding initial NRRt values) are shown in Figure 15. This suggests that there is in certain circumstances an economic case for undertaking a leakage-driven mains renewal programme on the basis that it is more cost-effective than allowing the network to deteriorate and undertaking additional active leakage control.

For an LRMC of 50-60p/m3 saved (for leak detection and repair) Figure 15 indicates that mains which have current NRRt values exceeding about 60m3/km/d per year may be candidates for renewal. This is equivalent in this illustration to mains having an expected remaining service life of at most 35 years – in the worst 3-4% of mains. It should be noted that the above example is intended to illustrate the principles of an approach, and clearly the results will be different for each network and distribution of infrastructure age, material and performance.

The figures used are those typical of average network densities in the UK – less than about 20m per connection. The NRR-density relationship shown in Figure 9 shows a substantial reduction in NRR for rural networks, which may therefore result in higher LRMC values for leakage-driven mains renewals. However, this may be offset to a degree by lower unit replacement costs. There is also evidence to show that the LRMC for find and fix is higher in rural networks. Further work is required to determine how the economic balance between mains renewal and find and fix changes in networks of different densities.


The rates of mains renewal proposed by some water undertakings in AMP4, if properly targeted, would appear to provide a means of at least maintaining current average natural rates of rise in leakage.

A framework has been demonstrated for comparing the long-run marginal costs of active leakage control and leakage-driven mains renewal based on an analysis of natural rates of rise in leakage.

An indication of typical NRR values where mains renewal can be an economic alternative to active leakage control has been provided. There are several data issues that would need to be addressed to support the proposed analysis framework and to ensure that the correct mains are being targeted for renewal:

  • The residual life of the mains infrastructure
  • The spatial distribution of NRR and background leakage
  • Forecasts of the temporal changes in NRR and background leakage as assets deteriorate
  • Local, for example DMA, estimates of the marginal costs of active leakage control and impacts of asset deterioration

Each of the above probably represents an information gap for most water undertakings and decision support systems are needed to gather and process the necessary data. The planning and appraisal process also needs to include an awareness of wider issues:

  • Increased disruption and increases in network pressures (leading to increased leakage)
  • Uncertainties associated with the savings achievable from particular renewal techniques
  • The need for a continuation of active leakage control to address leakage from customer supply pipes

It is essential, therefore, that leakage-driven mains renewal is appraised, not in isolation but as part of an integrated planning process.

There is much to be done by our industry to improve data availability to support an analysis framework such as that described above and a need to refine our understanding of some of the fundamental mechanisms on which leakage economic analysis relies so that decisions on possible policy shifts can be taken with confidence and ensure that past successes in leakage control are not put at risk.


This article is based on a presentation at Infrastructure Asset Maintenance & Management 11 January 2006. The author wishes to thank colleagues in RPS Water for their help and advice in its preparation.


1. Ofwat, Security of supply, leakage and the efficient use of water, 2004-05 Report

2. UKWIR, Capital Maintenance Planning: A Common Framework, 2002

3. Dellow, UKWIR Leakage Projects, 6th Annual Leakage Conference, Water UK, October 2005

4. WRc, UK Water Industry, Managing Leakage, 1994

5. Pearson, Fantozzi, Soares and Waldron, Searching for N2, 6th Annual Leakage Conference, Water UK, October 2005

6. DEFRA, Environment Agency, Ofwat, Leakage Target Setting for Water Companies in England and Wales, 2002

7. Parker, JM, Leakage and the Link to Asset Management, Leakage 2005, Halifax, Nova Scotia, September 2005

Action inspires action. Stay ahead of the curve with sustainability and energy newsletters from edie