Water scarcity is a harsh reality for many regions. As a result, reducing losses from water supply systems (WSSs) is of great environmental importance around the world. In France, water suppliers are legally bound to reduce losses from their WSSs through loss reduction action plans (LRAPs). For these plans to work, they need to suit the area to which they are applied. Their impacts on water bodies (WBs) and the environment also need to be taken into account. This paper explains an innovative approach to fulfil these objectives. It involves adding two elements to the design of LRAPs: calculation of a water abstraction and discharge balance (WADB) and analysis of environmental effects (AEE) relating to loss reduction actions, based on life cycle assessment. The usefulness and the practicality of this approach are examined through two practical case studies. It is shown that the WADB allows LRAPs to be configured to optimise water savings in the most sensitive WBs. AEE makes it possible to identify a loss reduction level above which the overall environmental balance becomes negative. While the LRAP design and the WADB calculation methods are ready to be used by WSS managers, the AEE still requires further improvement.
From a social and economic standpoint, reducing water losses in drinking water distribution networks is very important (Kingdom et al. 2006; Farley et al. 2008). In many countries, it is also crucial from an environmental point of view, because of water scarcity (Lehner et al. 2006).
In 2010, to address the challenge of reducing water wastage, the French State introduced new legislation governing water loss reduction. These rules require operators of drinking water supply systems (WSSs) to implement loss reduction action plans (LRAPs) when the efficiency of the system (E, ratio between current annual authorised consumption and annual system input volume) drops below a given threshold: between 65% and 85%, depending on the characteristics of the network and the scarcity of abstracted water resources (Ecology 2010).
Such a policy throws up a number of questions, the answers to which can vary greatly depending on the individual characteristics of each WSS: (i) Which actions are the most effective in reducing water losses? (ii) Given that a proportion of water lost from networks will return to the source from which it was abstracted, what is the true impact of loss reduction on water resource savings? (iii) How severe are the environmental impacts of actions taken to reduce losses, when compared with the environmental benefits of water savings?
To address these questions, ONEMA (‘Office National de l'Eau et des Milieux Aquatiques’, national agency for water) and the French Ministry for Ecology asked Irstea (‘Institut national de Recherche en Sciences et Technologies pour l'Environnement et l'Agriculture’, a French public research institute) developed a specific approach combining different perspectives and scales of water savings to optimize LRAPs.
MATERIAL AND METHODS
The approach aims to adjust LRAPs to optimize reduction of water abstraction and limit environmental impacts.
Three different methods were developed: (1) guide to creating a LRAP tailored to the characteristics of a given WSS, (2) water abstraction and discharge balance (WADB) for WSSs, and (3) analysis of environmental effects (AEE) relating to LRAPs.
Using the WADB and AEE, it is possible to evaluate the impacts of a given LRAP on water bodies (WBs) and the environment. By comparing different scenarios, LRAPs can be devised to increase water savings in sensitive areas, while at the same time limiting environmental impacts (Figure 1).
Designing an LRAP
The method for designing LRAPs is based on international work in the field of water loss reduction. The originality of the method lies in its use of performance indicators (PIs) imposed by French regulations, as well as being suited to the diverse sizes and levels of expertise characterising French water services. It involves four stages.
The first step is a preliminary diagnosis of the WSS, to gain a full understanding of its current status. This can generally be done in-house, using data already available.
The second stage involves improving calculation of PIs, as well as implementing immediate water loss reduction measures that do not require further investigation. These include improving or installing systems to measure water volumes, identifying and stopping visible water losses, targeted leak detection, and initial sectorization based on the hydraulic configuration of the network.
The third stage is an in-depth network study. This includes the creation of up-to-date network maps and inventories, the use of a hydraulic model, demand and resource analysis, sectorization plans, a measurement campaign, ranking sectors according to loss levels, and a historical analysis of network faults.
The final stage is to create a detailed action plan for the network. Thirty-eight possible measures were identified using a wide variety of technical publications from several countries, including work by (Farley 2001; AEAG, SMEGREG, OIEau 2005; Hamilton & Charamboulos 2013). To choose the best measures to apply to a given WSS, a decision tree was developed, inspired by the decision support tool created for the European Waterloss Project (2G-MED09-445) (Kanakoudis et al. 2015). It is based on values of indicators compared to pre-defined decision thresholds and responses to binary questions. Variables and indicators are either chosen from those required by French regulations or taken from existing studies and IWA guidelines (Lambert et al. 1999; Alegre et al. 2006). The decision tree is designed to be applied annually and allows selection and prioritization of actions on a sector-by-sector basis.
The method for creating LRAPs is detailed in a two-volume guide. The first volume (Renaud et al. 2014) includes two parts: Part 1 develops the step-by-step method, while Part 2 is made up of 38 factsheets describing individual loss reduction actions. The second volume (Aubrun et al. 2015) details the decision tree.
Adjusting the LRAP using a WADB
A WADB aims to identify how much of the water abstracted from WBs and fed into WSSs returns to those same WBs after use, thus establishing the volume of water saved in each WB through the application of LRAPs.
The suggested method by which loss reduction strategies can be designed to optimise water resource savings is as follows:
Identify the WBs from which water is abstracted and characterise the way in which they are recharged.
Identify all volumes of water in the WSS (consumption, losses, service water, etc.) and determine what proportion of those volumes is discharged through runoff, infiltration, or evapotranspiration.
Identify areas of runoff and infiltration and compare them with recharge sites of different WBs.
Generate a WADB for each WB used by the system.
Create loss reduction scenarios and generate a separate WADB for each of those scenarios.
Analyse the results and create an optimised action plan specifically suited to the status of the WBs concerned (i.e. applying the right actions in the right places to maximise water savings).
Adjusting the LRAP using an AEE
The aim of the AEE is to compare the adverse environmental effects of loss reduction activities with the environmental benefits gleaned through water savings (reduction of water production, treatment, and supply related impacts).
AEE was carried out using life cycle assessment (LCA). The method used (briefly described below) is detailed in (Pillot et al. 2016). LCA is a multi-criteria environmental impact assessment method used to quantify the potential impacts of human activities on ecosystems, natural resources, and human health. It involves assessing the impacts of natural resources being used, as well as environmental releases into air, water, and soil caused by the object studied during its life cycle. The LCA framework includes four methodological stages (ISO 2006): (i) goal and scope definition, (ii) life cycle inventory (LCI) analysis, (iii) life cycle impact assessment (LCIA), and (iv) interpretation of results.
The goal of AEE is to assess and compare the environmental impacts of water production with those caused by water loss reduction activities, notably as a result of the work, equipment, and infrastructures used. LCA is a relative approach, structured around a functional unit (FU). All subsequent analyses are linked to the chosen FU ‘avoided production of 1 m3 of drinking water’.
LCI involves listing all interactions between the environment and the boundaries of the system being studied. Suppliers’ technical documentation was used to create an inventory of raw materials, equipment, transport tools and energy consumption involved in loss reduction activities. Each item in the inventory was then converted into natural resource consumption and/or emissions of pollutants into the air, water or soils using the EcoInvent 3.1 international database (Frischknecht et al. 2005).
LCIA involves converting resources consumed and pollutants emitted into environmental impact indicators. This was achieved using the ReCiPe v1.11 method (Goedkoop et al. 2009). This method provides 18 categories of ‘midpoint’ impacts, as well as standardised aggregation of results into three ‘endpoint’ indicators that represent damage to human health, biodiversity and natural resources.
RESULTS AND DISCUSSION
The methods developed were applied to two study areas. Evaluation of the impacts of LRAPs on WBs (using a WADB) was tested on the WSS serving CABM (CABM, Communauté d'Agglomération de Béziers Méditérannée) in southeast France (Fisnot 2015). This site was chosen because of the diverse nature of its water resources, as well as a significant problem of over-abstraction in some WBs.
The AEE was carried out for the La Réole WSS in southwest France (Pillot et al. 2016). This WSS is representative of many small rural networks in France. Thanks to a number of previous studies carried out in the area, there is full access to existing data, and the network operator is very cooperative.
CABM case study
The CABM WSS provides water to around 110,000 permanent residents. A substantial amount of data provided by CABM and their main network operator, Suez was used to construct the WADB. These data included operational data from drinking and wastewater services, client information, geographical information systems, and hydraulic models. Based on these elements, it was possible to use the model developed by (Allaoui 2014) to identify, locate, and quantify system discharges (Figure 3).
Over half the total volume of water fed into the network is discharged through runoff, mainly wastewater. Infiltration also accounts for a large proportion of discharge, mainly through leaks, but also through garden watering and (to a lesser extent) on-site sewage disposal systems. Evapotranspiration (gardens and swimming pools) accounts for a relatively small proportion of discharge. While the destination of exported water is unknown, such water does not represent a significant proportion of discharge.
CABM abstracts a total of 11,000,000 m3 of water each year from four groundwater bodies, three of which have a ‘poor’ WFD (Water Framework Directive) status in terms of quantity, meaning that the withdrawal rate (including WSSs but also for other uses) is higher than the natural recharge. The two most critical are WB2 followed by WB4. The WADB showed very different water recharge rates (RR=RV/GAV) ranging from 0% for WB2 (deep groundwater body with its recharge site located far away), to more than 2,000% for WB3, a shallow groundwater body used very little by the CABM and whose recharge site covers a part of the WSS area (Table 1).
|.||WB 1 .||WB 2 .||WB 3 .||WB 4 .|
|Name||Hérault Alluvium||Astiens Sands of Valras Agde||Tertiary and Cretaceous||Downstream Orb Alluvium|
|Quantitative WFD state||Poor||Poor||Good||Poor|
|.||WB 1 .||WB 2 .||WB 3 .||WB 4 .|
|Name||Hérault Alluvium||Astiens Sands of Valras Agde||Tertiary and Cretaceous||Downstream Orb Alluvium|
|Quantitative WFD state||Poor||Poor||Good||Poor|
Three LRAP scenarios with the same gross saved volume (GSV) (173,000 m3) were designed using the decision tree method:
Scenario 1 ‘regulatory leak reduction’: Leak reduction activities confined to municipalities within the CABM whose loss levels exceeded the regulatory threshold.
Scenario 2 ‘homogeneous leak reduction’: Same level of leak reduction activity for all municipalities within the CABM.
Scenario 3 ‘targeted leak reduction’: Leak reduction activities focused on municipalities extracting water from the most at-risk water body (WB2).
Figure 4 shows that the net saved volume (NSV) for each WB is heavily dependent on the strategies applied.
This shows that a LRAP constructed using a usual method based on a leakage target (Trow & Farley 2004) (Scenario 1) is not the most effective way of saving water in WB2, and even exacerbates the existing deficit in WB4 (negative NSV). In Scenario 2, where leak reduction activity is uniform throughout the study area, water savings are achieved for WB4. However, the actual savings achieved (NSV) are lower than apparent savings (GSV) because the recharged volume (RV) is reduced. As expected, Scenario 3 leads to water savings for WB2. However, it also exacerbates the situation for WB4.
Using these results, it was possible to create an LRAP designed to reduce abstraction from priority WBs WB2 and WB4 by combining Scenarios 2 and 3.
La Réole case study
The La Réole WSS supplies 4,000 subscribers, abstracting in the process some 650,000 m3 of water. The groundwater body from which this water is abstracted is considered ‘poor’ in terms of quantity under the WFD.
Our study focuses on LRAPs implemented over a period of five years, made up of six commonly used actions: ‘network sectorization’, ‘assessing unmetered consumption’, ‘logging network interventions’, ‘pressure management’, ‘active leakage control using acoustic methods’ and ‘repairing leaking pipes’.
We modelled four consecutive LRAP scenarios leading to a gradual increase in efficiency for the WSS studied:
Actual observed situation, from 54% to 62%
First legal threshold, from 62% to 66.5%
Second legal threshold, from 66.5% to 71%
Maximum technical efficiency, from 71 to 77%.
The type and quantity of actions for each scenario were adjusted to achieve the desired efficiency. To estimate the volume of losses avoided by reducing pressure, we applied a commonly used equation, assuming proportionality between flow loss and pressure (Thornton 2003). The volume saved by detected leak repairs was considered to be a function of the intensity of leak detection activities, the effectiveness of those actions (considered linked to the length of district metered areas), and the estimated unitary annual volume of an undetected leak, taken as equal to 980 m3 according to average figures observed in the La Réole WSS, three times more than the ‘unavoidable volume’ suggested by (Lambert et al. 1999).
A LCI was carried out for each scenario. Data sources used in this part of the inventory were primary technical information from suppliers, field studies and expert analysis. Elementary components were constructed using existing material, energy, processes, transports and equipment from the international database Ecoinvent v3.1.
To remain as generic as possible, the LCI for the production of 1 m3 was not based on the real water production inventory from La Réole WSS, but on a European average taken from the Ecoinvent v3.1 database. The ReCiPe v1.11 method was then applied to evaluate LCIA.
To compare the impacts of the scenarios using the FU, we calculated the total avoided loss volume for each LRAP scenario over five years. Scenario 1: 412,000 m3; Scenario 2: 135,000 m3; Scenario 3: 155,000 m3; and Scenario 4: 148,000 m3.
The results of the LCA (Figure 5) show that the environmental benefits of an LRAP tend to reduce as network efficiency increases. For Scenario 1, the environmental benefits achieved through water savings are much higher than negative impacts due to work carried out to reduce leaks, but for Scenarios 2 and 3, impacts for ‘resources’ are higher than benefits, and for Scenario 4 it is also the case for ‘human health’.
CONCLUSION AND PERSPECTIVES
An approach was proposed to optimize LRAP and ensure they actually save water resources while also protecting the environment. The approach involves the following:
Using a decision tree to create WSS-specific LRAPs.
Calculating a WADB to estimate what proportion of the water abstracted from a given WB actually returns to that body.
Performing an AEE, based on LCA, to compare the environmental benefits and impacts of LRAPs.
The two case studies show that this approach can be applied to real situations. They also highlight the fact that LRAPs with the sole aim of increasing network efficiency are not necessarily the most effective in achieving water resource savings. It was also found that once water losses are brought down below a given level, the adverse environmental effects of LRAPs can outweigh any advantages achieved through water savings. This shows that in the same way that an ‘economical level of leakage’ (Howarth 1998), an ‘environmental level of leakage’ can also be calculated.
Using the WADB and AEE methods, it is possible to assess the effectiveness of a LRAP not only in terms of non-revenue water reduction but also from the point of view of impacts on water resources and the environment. In other words, the method answers two key questions:
Does water loss reduction from WSSs actually save water?
Is water loss reduction from WSSs beneficial for the environment?
The method of developing LRAPs is designed for a specific French context, and would need to be adapted for use elsewhere (notably in terms of PIs and thresholds).
The WADB and AEE methods are suitable for use in any country, although there would need to be some further improvements before they can be widely used in practice:
WADB is calculated for a calendar year at WB scale. However, water deficits can be seasonal in nature, with some WBs experiencing chronic shortages at certain times of the year. Also, water may move between different bodies (e.g. between rivers and associated aquifers).
AEE was carried out for a very specific case, and included a limited group of leak reduction actions. The environmental impacts of water production were calculated based on average values taken from a database. If this method is to be used in other contexts, it will be necessary to examine other leak reduction activities not covered in the case study, as well as developing a way to evaluate environmental impacts of water production on the specific area in which a given WSS is located (Kounina et al. 2013). On a broader note, there is much room for debate on the accuracy of methods used to calculate the water savings achieved through LRAPs, which adds significant uncertainty to the results of our AEEs.