Water Losses Analysis in Selected Group Water Supply Systems

Water losses are undesirable but inherent phenomenon occurring during exploitation process of all water supply systems [1]. The general definition of water losses is “the difference between water pumped into system and billed water” [2], with two general types: real losses and apparent losses. The amount of lost water is a parameter which individually characterizes each water supply system and that amount may be a few, over a dozen and even several dozen percent of a pumped water. Despite the fact that there are no legal standards of an acceptable water losses level, water companies endeavour to limit water losses as it directly minimizes water company costs. Water losses reduction reflects also in water quality improvement [3] and increase of water system reliability [4, 5]. However, the aspiration to eliminate water losses completely, despite its environmental benefits, is economically unprofitable. Therefore, water companies aim to limit water losses to the economically reasonable level [6] because its further reduction would generate costs higher than profits from saved water. The estimated economically reasonable level of water losses is approx. 8–10% [7] or even 5–6% [8], depending on the source. So far, the water losses phenomenon is well known in aspects of their reasons, detection and reducing methods [1-3, 7, 9, 10]. However, despite more and more precise methods of detection and reducing water losses, their exact evaluation is still impossible and estimated numerical values have an approximate character [9]. Therefore, to reduce water losses more effectively it is needed to minimize these of their reasons which generate the biggest losses. According to data [2], 80% up to 100% of real water losses are caused by water leakages from pipes. The elimination of water leakages is difficult due to the fact that leakages are often the result of random failures or breakages, impossible to predict. Therefore, recently the water outflow after pipe failure has become an intense subject of research studies [11-13].

Water losses may have multiple possible reasons, differing in accordance to a water supply system. Therefore, there is a high need to individually analyze each water supply distribution system, in order to develop a water losses reduction strategy. Additionally, it is also recommended to develop a general evaluation method, universal for all water systems, in order to compare water losses from different systems. In Poland, the beginning of water losses analysis was the moment of marketization of water prices and reduction of water demand consumption, which significantly emphasized the importance of water losses [14]. The aim of this paper is to analyze and compare water losses in selected two group water supply systems. The obtained results lead to evaluate the condition of analyzed water supply systems and to suggest potential actions in order to minimize water losses, which potentially can cause water delivery improvement and increase of economical profits.

According to the methodology, the first part of investigations was to assess water balances for the both networks in question. The balance components, that were used to calculate water losses indicators, are presented in Tables 3 and 4. The System Input Volume tends to remain stable over the whole period 2010–2015 in the both systems, although a slight decrease in SIV is observed (8.8% less in 2015 than in 2010) in the system A, and some fluctuations of up to 10% occur in the system B. Three other balance components connected with water losses tend to decrease in the both network cases, but in the system B the decrease is more clear – 17–19% between the beginning and the end of the period in the system A and 63–68% in the system B. SIV in the system A is near 1.8 times higher than in the system B, and Water Losses, Real Losses and Non-Revenue Water are near 2.5 times higher in 2010 and near 6 in 2015.

Values of water losses performance indicators listed in the subsection 2.2., calculated to completely assess and compare water losses in the both analysed distribution systems, are presented in Fig. 1 and 2. The obtained results suggest better technical condition of the network B – all analysed indicators are much lower than their analogues for the network A and show clear decrease tendency in the whole period. As a result, while in 2010 indicators WLP, WLL, RLL and ILI were about 1.7-1.8 times lower in the system B than in the system A and indicators WL% and NRWL were lower about 1.4-1.5 times, in 2015 the first mentioned group of indicators was lower as much as 4.5 times and the two other indicators were lower near 4 times. Clear visible reduction of water losses in the system B is the result of activities undertaken by the corporation managing this system, e.g. in 2011, about 50% of customer meters were replaced by high measurement accuracy devices with radio frequency AMR and since that time about 100 customer meters per year have been replaced.

Figure 1.

Figure 2.

The next part of the analysis was to compare the calculated water losses performance indicators to literature data (tables 5 and 6). Indicators WLP, WL% and WLL (Tab. 5) were referred to the networks similar to analysed ones according to water network intensity indicator (WNII = 6.38-8.35 m³/(day·km) for networks analysed in [23] and 13.2 m³/(day·km) for a network analysed in [24]) or length (123 km [24] and 315 km [25]). In the network A, values of indicators WLP and WL% are clearly higher than literature data, whereas WLL is near 2 times higher than a value given in [23], close to a value given in [24] and much lower than a value given in [25]. However, it should be emphasized that WLL should be compared in networks being characterized by similar WNII only [23] and WNII for a network described in [25] is unknown. In the network B, values of indicators WLP, WL% and WLL are lower than all literature data.

The other analysed indicators (Tab. 6) were compared to literature data for similar water network systems [26–27], and also to limit values, below which level of real losses is considered to be low [19–22]. The average value of RLL over the period 2010–2015 in the system A occurred more than 2 times higher than literature data as well as significantly exceeded the limit value proposed for water networks in rural areas according to German Water Loss Regulations [21]. In the system B, RLL was lower both than literature data, and the mentioned above limit value. In the system A, the average value of ILI was higher than literature data for Polish rural networks [26] and lower than value given for French conditions [27]. Moreover, the average value of ILI was higher in comparison to a limit value for low level of Real Losses according to German Regulations (it was in the range of values for medium level) and lower than the rest limit values. In the system B, the average ILI was definitely lower in comparison to literature data and it was not only in the range of limit values for low level of Real Losses, suggesting very good technical condition of the network, but also much lower than 1 – this means that Real Losses are lower than Unavoidable Annual Real Losses (UARL) in this network. Values of ILI, similar to the system B, were obtained for another group water supply system in Poland, described in [28]. According to [29], ILI < 1 can be a result of a failure to meet requirements for the empirical formula to calculate UARL (UARL value is necessary to calculate ILI). The UARL formula was originally developed for networks with average pressure at least 20 m H₂O and density of connections in the range of 20–100 per km. The first change in requirements for the UARL formula was the raising of the lower limit of the average pressure to 25 m H₂O, adding the requirement for minimal number of connections equalling 5000 and giving up the upper limit of connections density. The result of the next change was reduction of minimal number of connections to 3000 as well as the total abandonment of any connections density limits. In the latest change, all requirements were presented in one formula: a sum of connections number and a multiplied by 20 network length in km should not exceed 3000 [29]. The networks A and B meet the requirements according to latest change only, without any limitation of connections density, and this can cause unreliable ILI calculation results. However, three conditions according to the first change are still often used requirements for the UARL formula, especially in Poland [e.g. 9]. The networks A and B meet only one of these three conditions – average pressure head exceeds 25 m H₂O.

The last of indicators presented in Table 6 – Non-Revenue Water – was converted from percent to dm³ per connection per day (NRWcon) to be compared to a limit value according to [22]. Both in the system A and B, the average NRWcon over the period 2010–2015 exceeds limit value (75 dm³/(conn·day)) indicating very good network management [22]. As far as NRWcon values in consecutive years are concerned, in the system A all values were in the range 211.16–254.88 dm³/(conn·day) indicating existence of potential to improve the network management, whereas in the system B values of 44.04–61.78 dm³/(conn·day) in the second half of the analysed period let claim that the system is very well managed. Thus, the system A needs analysis to identify possibilities of Non-Revenue Water reduction, while in the system B potential for further Non-Revenue Water reductions is rather small.

The analysis of water losses in the water distribution systems A and B indicated better technical condition of the second system and suggests that this system is managed properly. The activities of the system B operator, aiming in reduction of water losses, enabled to achieve the desired objectives what is visible in significant decrease of all analyzed water losses indicators. In the system A, a decreasing tendency of the indicators over the period 2010–2015 indicates that the system operator also aims to reduce water losses, however, indicators values suggest that the undertaken activities are not sufficient. According to recommendations for networks being characterized by 150 < NRWcon < 300 dm³/(conn·day) [22], in the system A, customer meter management should be improved, possibility of pressure reduction should be analysed and active leakage control practices should be more efficient. Moreover, it is recommended to consider water flow meter placement in order to establish a water balance basing on the possible lowest number of estimated components.

Among the obtained results, low values of ILI may be a cause of doubts, especially in the case of the system B, for which in the second part of the period 2010–2015 ILI was much lower than 1, suggesting that Real Losses in the network B were much lower than unavoidable losses. The analyses indicate that softening requirements for UARL formula may be not adequate for condition of Polish rural networks. Another problem is overestimating a value of UARL by using the whole length of connections in calculation, because in Polish conditions information about the length of part of connection from a property line to a customer meter is usually unavailable. Taking into account the fact, that water distribution systems in rural areas are more than 76% of total length of all water distribution systems in Poland [16] and density of connections in these systems is often less than 20 per km, it seems to be sensible to put attention on necessity of developing the reliable method for UARL in rural networks.