The International Water Association water balance is an approach applied worldwide for determining and analysing water losses in water distribution systems (WDS) up to the point of customer metering. Thus, water losses occurring ‘before’ a customer meter are at the expenses of the water utility while water lost or wasted ‘after’ the meter is paid for by the customer. This applies to systems where customer metering is in place and/or consumption is charged according to the consumed volumes. However, many WDS in the world lack customer meters, are operated intermittently and a considerable amount of water is lost or wasted within the private property, e.g. by overflows of private tanks. The flat-rate tariff applied might not cover this amount or part of the amount. Thus, actual consumption and wastage should be separately quantified or estimated with respect to the utility's water reduction measures and the calculation of revenue water. This paper presents a water balance approach adapted to WDS which are operated intermittently, lack customer metering and charge flat-rate tariffs as well as a methodology for establishing the balance. The application is demonstrated for a district metered area of the city of Tiruvannamalai, India.
Hence, the IWA water balance covers those parts of a WDS the water utility is responsible for, i.e. up to the point of customer metering. Water losses occurring ‘before’ a customer meter are considered to be at the expense of the water utility while water lost or wasted ‘after’ the customer meter is assumed to be paid for by the customer. This assumption is valid for systems where customer metering is in place and/or water consumption is charged according to the consumed volumes. However, many WDS in the world, especially in developing countries, lack customer meters, are operated intermittently and apply a flat-rate tariff not reflecting the actual water consumption (Farley & Trow 2003; Kingdom et al. 2006; Fanner 2009; Klingel 2011). Furthermore, in intermittent supplies a considerable amount of water is lost or wasted within the customers' premises, e.g. by overflows of private tanks during supply or tank emptying prior to the supply period in order to get fresh water (Bradley et al. 2000; Totsuka et al. 2004; Butler & Memon 2006; Klingel 2011). Following the IWA water balance definition, all the water supplied to the customers is considered to be paid for by the customers. However, these assumptions do not hold in intermittent systems (Kumpel 2014). The flat-rate tariff actually might not cover the total amount of the supplied volume or part of it. Thus, actual consumption and wastage should be separately quantified or estimated with respect to the utility's water reduction measures and the calculation of revenue water.
Considering the constraints described above leads to the necessity of adapting the IWA water balance and the methods to determine the balance components. For the case of South Africa, Seago & Mckenzie (2007) and later McKenzie et al. (2012) subdivided the revenue water component of the IWA water balance. Subsidised water volumes are defined as free basic water, billed and actually paid volumes are called recovered revenue water and non-recovered revenue water is billed but not paid (Figure 1). Kanakoudis & Tsitsifli (2010) further modified that approach for the case of Greece by introducing the minimum charge difference. Customers pay a minimum charge equal to a certain minimum volume. The component comprises the proportion of the paid minimum volume actually not consumed (Figure 1). However, while the two modifications provide approaches for a detailed analysis of the revenue water component, intermittent supply as well as losses and wastage within the private property are still not considered.
This paper presents an adapted water balance for systems which are intermittently operated, apply a flat-rate tariff without customer metering and thus lack the preconditions to establish the standard IWA water balance. The developed approach enables a target-oriented assessment of water losses for WDS described above, as it features some components which are the most relevant, e.g. the wastage within the customers' premises. The method to determine the balance components is adapted and comprises metering of the supply to and the consumption within a sample of households as well as the statistical analysis of the sampled values.
First the adapted water balance is introduced before the methodology for determining the balance components is presented in the following section. Next, the application of the approach is demonstrated for a DMA of the city of Tiruvannamalai, Tamil Nadu, India. Finally a conclusion is given in the last section.
ADAPTED WATER BALANCE
Wastage within the private properties is authorised and billed due to the flat-rate tariff. However, these volumes should be quantified or at least estimated for determining the water loss reduction potential and checking the flat-rate tariff arrangement. Therefore, the Flat-Rate Billed Authorised Supply QBS is split into the two components Consumption QBSC and Wastage QBSW. The components System Input Volume QSIV, Water Losses QL and Real Losses QRL remain the same as in the IWA water balance as, for example, the type of losses stay the same in intermittently operated WDS. Apparent Losses QAL comprise only the sub-component Unauthorised Consumption QALC. The other two sub-components metering inaccuracies and systematic data handling errors from the IWA water balance are not depicted separately in the adapted water balance, as water meters are not installed and thus, billing is not based on meter readings. Nevertheless, known meter inaccuracies and systematic data handling errors of the monitoring system should already be considered when quantifying the volumes of System Input Volume QSIV as well as Flat-Rate Billed Authorised Supply QBS (outlined in the following sections).
The determination of Revenue Water QRW and Non-Revenue Water QNRW is not straightforward and cannot be calculated following the top down approach of the IWA water balance outlined in the introduction. This is due to the flat-rate tariff, which does not necessarily reflect the volume actually supplied to the customers. Therefore, both components have to be calculated separately, where three scenarios are possible. In scenario 1, customers actually are supplied with exactly the water volume that is billed. In this case Revenue Water QRW equals Flat-Rate Billed Authorised Supply QBS (QRW = QBS). In scenario 2, customers pay for more water than they are supplied with, thus, Revenue Water QRW is greater than Flat-Rate Billed Authorised Supply QBS (QRW > QBS). The amount of water consumers are not supplied with but pay for is defined as Excess Revenue Water QERW (QERW = QRW – QBS). In the third scenario, consumers pay for less water than they are supplied with. Hence, Revenue Water QRW is smaller than Flat-Rate Billed Authorised Supply QBS (QRW < QBS). The supplied water volumes assumed to be billed but actually not paid for by the consumers is called Billed Non-Revenue Water QBNRW (QBNRW = QBS – QRW).
Besides quantifying the components which are of particular interest regarding the technical aspects of the WDS, this approach allows for analysing whether the customers are billed exactly the supplied water, more than actually supplied or less. Regarding water loss reduction the focus should be put on Water Losses QL and Wastage QBSW instead of Non-Revenue Water QNRW. Especially in the case of scenario 2, in which Revenue Water QRW can be ‘artificially’ increased by charging for more water than actually supplied (QERW > 0). Furthermore, it has to be noted that often the revenue generated by the flat-rate tariff does not cover the cost for production and delivery of the Flat-Rate Volume QFR (Katko 1990; Le Blanc 2007).
DETERMINING THE WATER BALANCE COMPONENTS
System Input Volume QSIV
System Input Volume QSIV is defined as in the IWA water balance. The component should be measured precisely and comprehensively as well as be corrected for known meter inaccuracies.
Flat-Rate Billed Authorised Supply QBS
A full coverage of the initially unmetered WDS or DMA usually is not possible due to financial, social and/or political reasons. Hence, a representative and random sample of households is equipped with meters at the service connection to estimate Flat-Rate Billed Authorised Supply QBS by application of statistical sampling methods. The size of the sample cannot be accurately calculated due to the explorative nature of the method. However, 5% of the total number of households is assumed to ensure an adequate representativeness (Heller, personal communication).
It has to be noted that the metered values are afflicted with inaccuracies. Especially, the meters installed at the service connection are affected by the intermittent supply, e.g. by air intruded into the pipelines, see, for example, Van Zyl (2011). Therefore, possible customer metering inaccuracies have to be considered when estimating the Flat-Rate Billed Authorised Supply QBS, as this component comprises the water volume actually supplied to the customers. The obtained meter values should be corrected accordingly in advance of the statistical extrapolation using, for example, the approaches of Walter (2015) and Staiger (2016) regarding meter over-registration due to air flow.
Consumption QBSC and Wastage QBSW
As outlined above, a flat-rate water tariff without billing the actual Consumption QBSC could lead to increased Wastage QBSW within the customers' premises as there is no incentive to prevent wastage. In order to quantify how much of the supplied volume is actually consumed within the premises, a second meter could be installed. A typical household installation consists of one or more private storage tanks to cope with intermittent water supply (Galaitsi et al. 2016). In case the system pressure is sufficient, a roof tank is directly filled via the service connection of the household. If the system pressure is too low to directly fill the roof tank, a ground tank is installed which is supplied via the service connection. From this ground tank, the roof tank is filled by running a pump. Usually most of the Consumption QBSC of the household is covered by the roof tank, which provides an adequate water pressure at the taps within the household's premises (Roof Tank Consumption QRTC). Thus, the second meter is installed at the outlet of the roof tank to measure the Roof Tank Consumption QRTC,S at the sample households.
However, part of the supplied water might be consumed directly from the ground tank or anywhere else upstream of the roof tank (Ground Tank Consumption QGTC). Ground Tank Consumption QGTC,S of the sample can be estimated by a detailed survey of the daily consumption pattern of the customers, e.g. asking the customers to note the approximate volume of ground tank consumption in a protocol over a certain survey period.
It has to be noted that some households might store the water in a ground tank and directly pump the stored water to the taps inside the house and not to a roof tank. In this case the second meter should be installed directly downstream from the ground tank. Then this measured consumption has to be taken into account instead of the Roof Tank Consumption QRTC,S in Equation (4). The further steps to calculate the components Wastage QBSW and Consumption QBSC remain the same.
Unbilled Authorised Supply QUS
Apparent Losses QAL
As outlined before, QAL comprises only the volume of Unauthorised Consumption QALC which can be determined using state of the art methods, for example house-to-house surveys or estimations. Known meter inaccuracies and systematic data handling errors are already considered in the determination of System Input Volume QSIV and Flat-Rate Billed Authorised Supply QBS.
Real Losses QRL
Revenue Water QRW and Non-Revenue Water QNRW
CASE STUDY TIRUVANNAMALAI
For estimating the actual Flat-Rate Billed Authorised Supply QBS, customer meters (Zenner single-jet dry-running meters) were installed at a sample of 42 service connections. The locations of the households with metered service connections are marked in grey in Figure 3. With 42 of 710 service connections being metered, coverage of approximately 5.9% was achieved which is more than the recommended 5%. The selection was based on a preliminary household survey regarding household population and type of water supply as well as in consideration of the sample covering the whole DMA. The metered customers have been informed to be still charged as unmeasured and that the metered volumes are solely used for the research project and not for the billing process of the utility. Thereby, no changes in their consumption patterns should be ensured.
With the per capita supply data of the sample, the Flat-Rate Billed Authorised Supply QBS was estimated for the balancing period using the two statistical methods of simple sampling and post-stratified sampling, see Equations (1)–(3). Comparing the results of the two methods the post-stratification sampling method leads to a lower precision than the simple sampling method. The reason for that are the considerably high variances of the supply values within the strata, which impede the desired stratification effect. As a result, the simple sampling method using per capita supply values showed the highest precision, with an estimated Flat-Rate Billed Authorised Supply QBS of approximately 13,035 m3 for the balance period of 23 days. Subtracting inaccuracies of the customer meters due to air flow applying the approach of Walter (2015) and Staiger (2016) results in a corrected Flat-Rate Billed Authorised Supply QBS of approximately 11,613 m3 (75%).
For estimating the Consumption QBSC and Wastage QBSW, additional customer meters have been installed downstream of the roof tanks at the selected household sample. Two household surveys with the customers taking notes of their approximate ground tank consumption were executed. The surveyed customers have been informed that the data is solely used for research purposes. Consumption QBSC of 7,504 m3 (49%) was calculated with Equation (4) using the survey results and the measurement data of the roof tank meters. With Equation (5), Wastage QBSW was estimated to be 4,109 m3 (26%).
Unbilled Authorised Supply QUS comprises 0 m3 as no volumes are taken by unbilled authorised customers, the utility or other authorised parties within the DMA. Hence, Water Losses QL equal approx. 3,787 m3 (25%) within the balancing period considering the above mentioned numbers (QL = QSIV – QBS – QUS). Apparent Losses QAL are 0 m3 as no Unauthorised Consumption QALC is reported in the balancing period. Thus, Real Losses QRL equal Water Losses QL (Equation (6)).
According to the flat-rate tariff charged in Tiruvannamalai, customers pay 100 Indian rupees (INR) for the supply of 7.5 m3 per month in addition to a constant water supply tax depending on the property size (Tiruvannamalai Municipality 2013). With QFR = 7.5 m3, NC = 710 and t = 0.76, Revenue Water QRW can be calculated for the balancing period to 4,083 m3 (26%) using Equation (7). Non-Revenue Water QNRW is 11,317 m3 (74%) with Equation (9). Thus, the water volume supplied to the customers is considerably higher than the volume billed to and paid by the customers charged with the flat-rate tariff (QBS > QRW, scenario 3). Accordingly, Equation (11) gives Billed Non-Revenue Water QBNRW = 7,530 m3. More than one third of Non-Revenue Water QNRW is due to Wastage QBSW within the customers' premises.
This paper presents the adaption of the IWA water balance to intermittently operated WDS applying a flat-rate tariff without metering and billing of the actual consumption. In consideration of these conditions, authorised consumption from the IWA water balance is replaced by the component Authorised Supply QS and its two components Flat-Rate Billed Authorised Supply QBS and Unbilled Authorised Supply QUS. The Flat-Rate Billed Authorised Supply QBS comprises the two components Consumption QBSC and Wastage QBSW, as part of the supplied water volume might be wasted due to overflowing private storage tanks or emptying of these tanks in order to get fresh water.
The methodologies to quantify or estimate the volumes of the above mentioned balance components also have to be adapted. In a WDS with initially unmetered customers, a monitoring system has to be implemented to measure Flat-Rate Billed Authorised Supply QBS, Consumption QBSC and Wastage QBSW of a sample of households with water meters installed upstream of the customers' premises and downstream of the customers' roof tank. Applying statistical methods such as simple sampling or post-stratified sampling along with auxiliary information of a comprehensive household survey, the volumes of Flat-Rate Billed Authorised Supply QBS, Consumption QBSC and Wastage QBSW of the total population can be estimated. In contrast to the IWA water balance, the deduction of the revenue and non-revenue water is not straightforward, as the customers are not billed according to their actual consumption but with a flat-rate tariff often related to a certain fixed (theoretical) water volume. Thus, the Revenue Water QRW might be higher, equal or smaller than the actual Flat-Rate Billed Authorised Supply QBS. All three scenarios are also considered and depicted in the adapted water balance.
The application of the outlined approach is demonstrated for a DMA with approximately 710 household service connections in the medium-sized city of Tiruvannamalai in South India. The implemented monitoring system to estimate Flat-Rate Billed Authorised Supply QBS, the Consumption QBSC and the Wastage QBSW covers a sample of 42 household service connections. The statistical methods applied to quantify the Flat-Rate Billed Authorised Supply QBS of the DMA show that additional information of household occupancy increases the precision of the estimate and should therefore be acquired by a household survey or census data. The results of the adapted water balance show the volume of Water Losses QL being about 25% and the volume of Wastage QBSW within the customers' premises about 26% of the System Input Volume QSIV. The volume of Non-Revenue Water QNRW is considerably high and about 74% of QSIV. Two-thirds of QNRW are due to the inadequate flat-rate tariff, as the utility charges far less volume than actually supplied to the customers.
The application of the adapted water balance in the case study shows that this approach can be used to initiate a continuous monitoring and assessment of the water losses and non-revenue water for WDS which charge a flat-rate tariff without metering and billing of the actual consumption. Based on the statistical analysis of supply and consumption volumes of a metered household sample, the adapted water balance focuses on the most relevant components for water losses in WDS described above. In a next step, developing indicators derived from the adapted water balance in addition to the internationally recognised and applied performance indicators, for example, introduced by the IWA and comprehensively described by Alegre et al. (2006), would be a valuable help for the water loss management in the given cases. Operational indicators, for example, could provide information whether water loss reduction measures should be concentrated on the network or the customers' premises along with financial indicators helping to interpret the subcomponents of Revenue Water QRW and Non-Revenue Water QNRW. With the adapted water balance and performance indicators, the developed approach could serve as a tool for many water utilities worldwide, especially in developing countries, to assess and tackle water losses in their distribution systems.
The authors are grateful to the German Federal Ministry of Education and Research for funding the work (funding code: 02WCL1300B).